[Qemu-devel] [PATCH v4 0/3] block/stream: get rid of the base

[Qemu-devel] [PATCH v4 0/3] block/stream: get rid of the base

[Andrey Shinkevich]

This series introduces a bottom intermediate node that eliminates the
dependency on the base that may change while stream job is running.
It happens when stream/commit parallel jobs are running on the same
backing chain. The base node of the stream job may be a top node of
the parallel commit job and can change before the stream job is
completed. We avoid that dependency by introducing the bottom node.

v4:
trace_stream_start reverted to the base.
bdrv_is_allocated_above_inclusive() deleted and the new parameter
'bool include_base' was added to the bdrv_is_allocated_above().

v3:
bottom = bdrv_find_overlay() moved into the stream_start()

v2:
The function bdrv_find_overlay() with the same functionality has been
used in the qmp_block_stream() to find a bottom node.
The algorithm in the bdrv_do_is_allocated_above() was simplified.
The commit messages were modified.

Discussed in the e-mail threads with the message IDs
<1553866154-257311-1-git-send-email-andrey.shinkevich@virtuozzo.com>
<1550762799-830661-1-git-send-email-andrey.shinkevich@virtuozzo.com>
<1554120365-39119-1-git-send-email-andrey.shinkevich@virtuozzo.com>
<1554483379-682051-1-git-send-email-andrey.shinkevich@virtuozzo.com>

Andrey Shinkevich (2):
  block: include base when checking image chain for block allocation
  block/stream: introduce a bottom node

Vladimir Sementsov-Ogievskiy (1):
  block/stream: refactor stream_run: drop goto

 block/commit.c         |  2 +-
 block/io.c             | 14 +++++++++---
 block/mirror.c         |  2 +-
 block/replication.c    |  2 +-
 block/stream.c         | 62 +++++++++++++++++++++++---------------------------
 include/block/block.h  |  3 ++-
 tests/qemu-iotests/245 |  4 ++--
 7 files changed, 47 insertions(+), 42 deletions(-)

-- 
1.8.3.1

[Qemu-devel] [PATCH v4 0/3] block/stream: get rid of the base

[Andrey Shinkevich]

This series introduces a bottom intermediate node that eliminates the
dependency on the base that may change while stream job is running.
It happens when stream/commit parallel jobs are running on the same
backing chain. The base node of the stream job may be a top node of
the parallel commit job and can change before the stream job is
completed. We avoid that dependency by introducing the bottom node.

v4:
trace_stream_start reverted to the base.
bdrv_is_allocated_above_inclusive() deleted and the new parameter
'bool include_base' was added to the bdrv_is_allocated_above().

v3:
bottom = bdrv_find_overlay() moved into the stream_start()

v2:
The function bdrv_find_overlay() with the same functionality has been
used in the qmp_block_stream() to find a bottom node.
The algorithm in the bdrv_do_is_allocated_above() was simplified.
The commit messages were modified.

Discussed in the e-mail threads with the message IDs
<1553866154-257311-1-git-send-email-andrey.shinkevich@virtuozzo.com>
<1550762799-830661-1-git-send-email-andrey.shinkevich@virtuozzo.com>
<1554120365-39119-1-git-send-email-andrey.shinkevich@virtuozzo.com>
<1554483379-682051-1-git-send-email-andrey.shinkevich@virtuozzo.com>

Andrey Shinkevich (2):
  block: include base when checking image chain for block allocation
  block/stream: introduce a bottom node

Vladimir Sementsov-Ogievskiy (1):
  block/stream: refactor stream_run: drop goto

 block/commit.c         |  2 +-
 block/io.c             | 14 +++++++++---
 block/mirror.c         |  2 +-
 block/replication.c    |  2 +-
 block/stream.c         | 62 +++++++++++++++++++++++---------------------------
 include/block/block.h  |  3 ++-
 tests/qemu-iotests/245 |  4 ++--
 7 files changed, 47 insertions(+), 42 deletions(-)

-- 
1.8.3.1

[Qemu-devel] [PATCH v4 1/3] block: include base when checking image chain for block allocation

[Andrey Shinkevich]

This patch is used in the 'block/stream: introduce a bottom node'
that is following. Instead of the base node, the caller may pass
the node that has the base as its backing image to the function
bdrv_is_allocated_above() with a new parameter include_base = true
and get rid of the dependency on the base that may change during
commit/stream parallel jobs. Now, if the specified base is not
found in the backing image chain, the QEMU will crash.

Suggested-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Signed-off-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
---
 block/commit.c        |  2 +-
 block/io.c            | 14 +++++++++++---
 block/mirror.c        |  2 +-
 block/replication.c   |  2 +-
 block/stream.c        |  2 +-
 include/block/block.h |  3 ++-
 6 files changed, 17 insertions(+), 8 deletions(-)

diff --git a/block/commit.c b/block/commit.c
index ba60fef..1bd5284 100644
--- a/block/commit.c
+++ b/block/commit.c
@@ -177,7 +177,7 @@ static int coroutine_fn commit_run(Job *job, Error **errp)
             break;
         }
         /* Copy if allocated above the base */
-        ret = bdrv_is_allocated_above(blk_bs(s->top), blk_bs(s->base),
+        ret = bdrv_is_allocated_above(blk_bs(s->top), blk_bs(s->base), false,
                                       offset, COMMIT_BUFFER_SIZE, &n);
         copy = (ret == 1);
         trace_commit_one_iteration(s, offset, n, ret);
diff --git a/block/io.c b/block/io.c
index dfc153b..1ed4164 100644
--- a/block/io.c
+++ b/block/io.c
@@ -2317,7 +2317,8 @@ int coroutine_fn bdrv_is_allocated(BlockDriverState *bs, int64_t offset,
  * Given an image chain: ... -> [BASE] -> [INTER1] -> [INTER2] -> [TOP]
  *
  * Return true if (a prefix of) the given range is allocated in any image
- * between BASE and TOP (inclusive).  BASE can be NULL to check if the given
+ * between BASE and TOP (TOP included). To check the BASE image, set the
+ * 'include_base' to 'true'. The BASE can be NULL to check if the given
  * offset is allocated in any image of the chain.  Return false otherwise,
  * or negative errno on failure.
  *
@@ -2331,14 +2332,17 @@ int coroutine_fn bdrv_is_allocated(BlockDriverState *bs, int64_t offset,
  */
 int bdrv_is_allocated_above(BlockDriverState *top,
                             BlockDriverState *base,
-                            int64_t offset, int64_t bytes, int64_t *pnum)
+                            bool include_base, int64_t offset,
+                            int64_t bytes, int64_t *pnum)
 {
     BlockDriverState *intermediate;
     int ret;
     int64_t n = bytes;
 
+    assert(base || !include_base);
+
     intermediate = top;
-    while (intermediate && intermediate != base) {
+    while (include_base || intermediate != base) {
         int64_t pnum_inter;
         int64_t size_inter;
 
@@ -2360,6 +2364,10 @@ int bdrv_is_allocated_above(BlockDriverState *top,
             n = pnum_inter;
         }
 
+        if (intermediate == base) {
+            break;
+        }
+
         intermediate = backing_bs(intermediate);
     }
 
diff --git a/block/mirror.c b/block/mirror.c
index ff15cfb..923548f 100644
--- a/block/mirror.c
+++ b/block/mirror.c
@@ -807,7 +807,7 @@ static int coroutine_fn mirror_dirty_init(MirrorBlockJob *s)
             return 0;
         }
 
-        ret = bdrv_is_allocated_above(bs, base, offset, bytes, &count);
+        ret = bdrv_is_allocated_above(bs, base, false, offset, bytes, &count);
         if (ret < 0) {
             return ret;
         }
diff --git a/block/replication.c b/block/replication.c
index 3d4dedd..fc8d2ad 100644
--- a/block/replication.c
+++ b/block/replication.c
@@ -272,7 +272,7 @@ static coroutine_fn int replication_co_writev(BlockDriverState *bs,
     while (remaining_sectors > 0) {
         int64_t count;
 
-        ret = bdrv_is_allocated_above(top->bs, base->bs,
+        ret = bdrv_is_allocated_above(top->bs, base->bs, false,
                                       sector_num * BDRV_SECTOR_SIZE,
                                       remaining_sectors * BDRV_SECTOR_SIZE,
                                       &count);
diff --git a/block/stream.c b/block/stream.c
index bfaebb8..0920092 100644
--- a/block/stream.c
+++ b/block/stream.c
@@ -162,7 +162,7 @@ static int coroutine_fn stream_run(Job *job, Error **errp)
         } else if (ret >= 0) {
             /* Copy if allocated in the intermediate images.  Limit to the
              * known-unallocated area [offset, offset+n*BDRV_SECTOR_SIZE).  */
-            ret = bdrv_is_allocated_above(backing_bs(bs), base,
+            ret = bdrv_is_allocated_above(backing_bs(bs), base, false,
                                           offset, n, &n);
 
             /* Finish early if end of backing file has been reached */
diff --git a/include/block/block.h b/include/block/block.h
index c7a2619..f98e858 100644
--- a/include/block/block.h
+++ b/include/block/block.h
@@ -447,7 +447,8 @@ int bdrv_block_status_above(BlockDriverState *bs, BlockDriverState *base,
 int bdrv_is_allocated(BlockDriverState *bs, int64_t offset, int64_t bytes,
                       int64_t *pnum);
 int bdrv_is_allocated_above(BlockDriverState *top, BlockDriverState *base,
-                            int64_t offset, int64_t bytes, int64_t *pnum);
+                            bool include_base, int64_t offset, int64_t bytes,
+                            int64_t *pnum);
 
 bool bdrv_is_read_only(BlockDriverState *bs);
 int bdrv_can_set_read_only(BlockDriverState *bs, bool read_only,
-- 
1.8.3.1

[Qemu-devel] [PATCH v4 1/3] block: include base when checking image chain for block allocation

[Andrey Shinkevich]

This patch is used in the 'block/stream: introduce a bottom node'
that is following. Instead of the base node, the caller may pass
the node that has the base as its backing image to the function
bdrv_is_allocated_above() with a new parameter include_base = true
and get rid of the dependency on the base that may change during
commit/stream parallel jobs. Now, if the specified base is not
found in the backing image chain, the QEMU will crash.

Suggested-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Signed-off-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
---
 block/commit.c        |  2 +-
 block/io.c            | 14 +++++++++++---
 block/mirror.c        |  2 +-
 block/replication.c   |  2 +-
 block/stream.c        |  2 +-
 include/block/block.h |  3 ++-
 6 files changed, 17 insertions(+), 8 deletions(-)

diff --git a/block/commit.c b/block/commit.c
index ba60fef..1bd5284 100644
--- a/block/commit.c
+++ b/block/commit.c
@@ -177,7 +177,7 @@ static int coroutine_fn commit_run(Job *job, Error **errp)
             break;
         }
         /* Copy if allocated above the base */
-        ret = bdrv_is_allocated_above(blk_bs(s->top), blk_bs(s->base),
+        ret = bdrv_is_allocated_above(blk_bs(s->top), blk_bs(s->base), false,
                                       offset, COMMIT_BUFFER_SIZE, &n);
         copy = (ret == 1);
         trace_commit_one_iteration(s, offset, n, ret);
diff --git a/block/io.c b/block/io.c
index dfc153b..1ed4164 100644
--- a/block/io.c
+++ b/block/io.c
@@ -2317,7 +2317,8 @@ int coroutine_fn bdrv_is_allocated(BlockDriverState *bs, int64_t offset,
  * Given an image chain: ... -> [BASE] -> [INTER1] -> [INTER2] -> [TOP]
  *
  * Return true if (a prefix of) the given range is allocated in any image
- * between BASE and TOP (inclusive).  BASE can be NULL to check if the given
+ * between BASE and TOP (TOP included). To check the BASE image, set the
+ * 'include_base' to 'true'. The BASE can be NULL to check if the given
  * offset is allocated in any image of the chain.  Return false otherwise,
  * or negative errno on failure.
  *
@@ -2331,14 +2332,17 @@ int coroutine_fn bdrv_is_allocated(BlockDriverState *bs, int64_t offset,
  */
 int bdrv_is_allocated_above(BlockDriverState *top,
                             BlockDriverState *base,
-                            int64_t offset, int64_t bytes, int64_t *pnum)
+                            bool include_base, int64_t offset,
+                            int64_t bytes, int64_t *pnum)
 {
     BlockDriverState *intermediate;
     int ret;
     int64_t n = bytes;
 
+    assert(base || !include_base);
+
     intermediate = top;
-    while (intermediate && intermediate != base) {
+    while (include_base || intermediate != base) {
         int64_t pnum_inter;
         int64_t size_inter;
 
@@ -2360,6 +2364,10 @@ int bdrv_is_allocated_above(BlockDriverState *top,
             n = pnum_inter;
         }
 
+        if (intermediate == base) {
+            break;
+        }
+
         intermediate = backing_bs(intermediate);
     }
 
diff --git a/block/mirror.c b/block/mirror.c
index ff15cfb..923548f 100644
--- a/block/mirror.c
+++ b/block/mirror.c
@@ -807,7 +807,7 @@ static int coroutine_fn mirror_dirty_init(MirrorBlockJob *s)
             return 0;
         }
 
-        ret = bdrv_is_allocated_above(bs, base, offset, bytes, &count);
+        ret = bdrv_is_allocated_above(bs, base, false, offset, bytes, &count);
         if (ret < 0) {
             return ret;
         }
diff --git a/block/replication.c b/block/replication.c
index 3d4dedd..fc8d2ad 100644
--- a/block/replication.c
+++ b/block/replication.c
@@ -272,7 +272,7 @@ static coroutine_fn int replication_co_writev(BlockDriverState *bs,
     while (remaining_sectors > 0) {
         int64_t count;
 
-        ret = bdrv_is_allocated_above(top->bs, base->bs,
+        ret = bdrv_is_allocated_above(top->bs, base->bs, false,
                                       sector_num * BDRV_SECTOR_SIZE,
                                       remaining_sectors * BDRV_SECTOR_SIZE,
                                       &count);
diff --git a/block/stream.c b/block/stream.c
index bfaebb8..0920092 100644
--- a/block/stream.c
+++ b/block/stream.c
@@ -162,7 +162,7 @@ static int coroutine_fn stream_run(Job *job, Error **errp)
         } else if (ret >= 0) {
             /* Copy if allocated in the intermediate images.  Limit to the
              * known-unallocated area [offset, offset+n*BDRV_SECTOR_SIZE).  */
-            ret = bdrv_is_allocated_above(backing_bs(bs), base,
+            ret = bdrv_is_allocated_above(backing_bs(bs), base, false,
                                           offset, n, &n);
 
             /* Finish early if end of backing file has been reached */
diff --git a/include/block/block.h b/include/block/block.h
index c7a2619..f98e858 100644
--- a/include/block/block.h
+++ b/include/block/block.h
@@ -447,7 +447,8 @@ int bdrv_block_status_above(BlockDriverState *bs, BlockDriverState *base,
 int bdrv_is_allocated(BlockDriverState *bs, int64_t offset, int64_t bytes,
                       int64_t *pnum);
 int bdrv_is_allocated_above(BlockDriverState *top, BlockDriverState *base,
-                            int64_t offset, int64_t bytes, int64_t *pnum);
+                            bool include_base, int64_t offset, int64_t bytes,
+                            int64_t *pnum);
 
 bool bdrv_is_read_only(BlockDriverState *bs);
 int bdrv_can_set_read_only(BlockDriverState *bs, bool read_only,
-- 
1.8.3.1

[Qemu-devel] [PATCH v4 2/3] block/stream: refactor stream_run: drop goto

[Andrey Shinkevich]

From: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>

The goto is unnecessary in the stream_run() since the common exit
code was removed in the commit eb23654dbe43b549ea2a9ebff9d8e:
"jobs: utilize job_exit shim".

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Signed-off-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
---
 block/stream.c | 13 ++++---------
 1 file changed, 4 insertions(+), 9 deletions(-)

diff --git a/block/stream.c b/block/stream.c
index 0920092..e73a320 100644
--- a/block/stream.c
+++ b/block/stream.c
@@ -122,13 +122,12 @@ static int coroutine_fn stream_run(Job *job, Error **errp)
     void *buf;
 
     if (!bs->backing) {
-        goto out;
+        return 0;
     }
 
     len = bdrv_getlength(bs);
     if (len < 0) {
-        ret = len;
-        goto out;
+        return len;
     }
     job_progress_set_remaining(&s->common.job, len);
 
@@ -205,14 +204,10 @@ static int coroutine_fn stream_run(Job *job, Error **errp)
         bdrv_disable_copy_on_read(bs);
     }
 
-    /* Do not remove the backing file if an error was there but ignored.  */
-    ret = error;
-
     qemu_vfree(buf);
 
-out:
-    /* Modify backing chain and close BDSes in main loop */
-    return ret;
+    /* Do not remove the backing file if an error was there but ignored. */
+    return error;
 }
 
 static const BlockJobDriver stream_job_driver = {
-- 
1.8.3.1

[Qemu-devel] [PATCH v4 2/3] block/stream: refactor stream_run: drop goto

[Andrey Shinkevich]

From: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>

The goto is unnecessary in the stream_run() since the common exit
code was removed in the commit eb23654dbe43b549ea2a9ebff9d8e:
"jobs: utilize job_exit shim".

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Signed-off-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
---
 block/stream.c | 13 ++++---------
 1 file changed, 4 insertions(+), 9 deletions(-)

diff --git a/block/stream.c b/block/stream.c
index 0920092..e73a320 100644
--- a/block/stream.c
+++ b/block/stream.c
@@ -122,13 +122,12 @@ static int coroutine_fn stream_run(Job *job, Error **errp)
     void *buf;
 
     if (!bs->backing) {
-        goto out;
+        return 0;
     }
 
     len = bdrv_getlength(bs);
     if (len < 0) {
-        ret = len;
-        goto out;
+        return len;
     }
     job_progress_set_remaining(&s->common.job, len);
 
@@ -205,14 +204,10 @@ static int coroutine_fn stream_run(Job *job, Error **errp)
         bdrv_disable_copy_on_read(bs);
     }
 
-    /* Do not remove the backing file if an error was there but ignored.  */
-    ret = error;
-
     qemu_vfree(buf);
 
-out:
-    /* Modify backing chain and close BDSes in main loop */
-    return ret;
+    /* Do not remove the backing file if an error was there but ignored. */
+    return error;
 }
 
 static const BlockJobDriver stream_job_driver = {
-- 
1.8.3.1

[Qemu-devel] [PATCH v4 3/3] block/stream: introduce a bottom node

[Andrey Shinkevich]

The bottom node is the intermediate block device that has the base as its
backing image. It is used instead of the base node while a block stream
job is running to avoid dependency on the base that may change due to the
parallel jobs. The change may take place due to a filter node as well that
is inserted between the base and the intermediate bottom node. It occurs
when the base node is the top one for another commit or stream job.
After the introduction of the bottom node, don't freeze its backing child,
that's the base, anymore.

Suggested-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Signed-off-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
---
 block/stream.c         | 49 +++++++++++++++++++++++++------------------------
 tests/qemu-iotests/245 |  4 ++--
 2 files changed, 27 insertions(+), 26 deletions(-)

diff --git a/block/stream.c b/block/stream.c
index e73a320..c39212e 100644
--- a/block/stream.c
+++ b/block/stream.c
@@ -31,7 +31,7 @@ enum {
 
 typedef struct StreamBlockJob {
     BlockJob common;
-    BlockDriverState *base;
+    BlockDriverState *bottom;
     BlockdevOnError on_error;
     char *backing_file_str;
     bool bs_read_only;
@@ -56,7 +56,7 @@ static void stream_abort(Job *job)
 
     if (s->chain_frozen) {
         BlockJob *bjob = &s->common;
-        bdrv_unfreeze_backing_chain(blk_bs(bjob->blk), s->base);
+        bdrv_unfreeze_backing_chain(blk_bs(bjob->blk), s->bottom);
     }
 }
 
@@ -65,11 +65,11 @@ static int stream_prepare(Job *job)
     StreamBlockJob *s = container_of(job, StreamBlockJob, common.job);
     BlockJob *bjob = &s->common;
     BlockDriverState *bs = blk_bs(bjob->blk);
-    BlockDriverState *base = s->base;
+    BlockDriverState *base = backing_bs(s->bottom);
     Error *local_err = NULL;
     int ret = 0;
 
-    bdrv_unfreeze_backing_chain(bs, base);
+    bdrv_unfreeze_backing_chain(bs, s->bottom);
     s->chain_frozen = false;
 
     if (bs->backing) {
@@ -112,7 +112,7 @@ static int coroutine_fn stream_run(Job *job, Error **errp)
     StreamBlockJob *s = container_of(job, StreamBlockJob, common.job);
     BlockBackend *blk = s->common.blk;
     BlockDriverState *bs = blk_bs(blk);
-    BlockDriverState *base = s->base;
+    bool enable_cor = !backing_bs(s->bottom);
     int64_t len;
     int64_t offset = 0;
     uint64_t delay_ns = 0;
@@ -121,7 +121,8 @@ static int coroutine_fn stream_run(Job *job, Error **errp)
     int64_t n = 0; /* bytes */
     void *buf;
 
-    if (!bs->backing) {
+    if (bs == s->bottom) {
+        /* Nothing to stream */
         return 0;
     }
 
@@ -138,7 +139,7 @@ static int coroutine_fn stream_run(Job *job, Error **errp)
      * backing chain since the copy-on-read operation does not take base into
      * account.
      */
-    if (!base) {
+    if (enable_cor) {
         bdrv_enable_copy_on_read(bs);
     }
 
@@ -161,9 +162,8 @@ static int coroutine_fn stream_run(Job *job, Error **errp)
         } else if (ret >= 0) {
             /* Copy if allocated in the intermediate images.  Limit to the
              * known-unallocated area [offset, offset+n*BDRV_SECTOR_SIZE).  */
-            ret = bdrv_is_allocated_above(backing_bs(bs), base, false,
+            ret = bdrv_is_allocated_above(backing_bs(bs), s->bottom, true,
                                           offset, n, &n);
-
             /* Finish early if end of backing file has been reached */
             if (ret == 0 && n == 0) {
                 n = len - offset;
@@ -200,7 +200,7 @@ static int coroutine_fn stream_run(Job *job, Error **errp)
         }
     }
 
-    if (!base) {
+    if (enable_cor) {
         bdrv_disable_copy_on_read(bs);
     }
 
@@ -232,8 +232,10 @@ void stream_start(const char *job_id, BlockDriverState *bs,
     StreamBlockJob *s;
     BlockDriverState *iter;
     bool bs_read_only;
+    int basic_flags = BLK_PERM_CONSISTENT_READ | BLK_PERM_WRITE_UNCHANGED;
+    BlockDriverState *bottom = bdrv_find_overlay(bs, base);
 
-    if (bdrv_freeze_backing_chain(bs, base, errp) < 0) {
+    if (bdrv_freeze_backing_chain(bs, bottom, errp) < 0) {
         return;
     }
 
@@ -250,26 +252,25 @@ void stream_start(const char *job_id, BlockDriverState *bs,
      * already have our own plans. Also don't allow resize as the image size is
      * queried only at the job start and then cached. */
     s = block_job_create(job_id, &stream_job_driver, NULL, bs,
-                         BLK_PERM_CONSISTENT_READ | BLK_PERM_WRITE_UNCHANGED |
-                         BLK_PERM_GRAPH_MOD,
-                         BLK_PERM_CONSISTENT_READ | BLK_PERM_WRITE_UNCHANGED |
-                         BLK_PERM_WRITE,
+                         basic_flags | BLK_PERM_GRAPH_MOD,
+                         basic_flags | BLK_PERM_WRITE,
                          speed, creation_flags, NULL, NULL, errp);
     if (!s) {
         goto fail;
     }
 
-    /* Block all intermediate nodes between bs and base, because they will
-     * disappear from the chain after this operation. The streaming job reads
-     * every block only once, assuming that it doesn't change, so block writes
-     * and resizes. */
-    for (iter = backing_bs(bs); iter && iter != base; iter = backing_bs(iter)) {
-        block_job_add_bdrv(&s->common, "intermediate node", iter, 0,
-                           BLK_PERM_CONSISTENT_READ | BLK_PERM_WRITE_UNCHANGED,
-                           &error_abort);
+    /*
+     * Block all intermediate nodes between bs and bottom (inclusive), because
+     * they will disappear from the chain after this operation. The streaming
+     * job reads every block only once, assuming that it doesn't change, so
+     * forbid writes and resizes.
+     */
+    for (iter = bs; iter != bottom; iter = backing_bs(iter)) {
+        block_job_add_bdrv(&s->common, "intermediate node", backing_bs(iter),
+                           0, basic_flags, &error_abort);
     }
 
-    s->base = base;
+    s->bottom = bottom;
     s->backing_file_str = g_strdup(backing_file_str);
     s->bs_read_only = bs_read_only;
     s->chain_frozen = true;
diff --git a/tests/qemu-iotests/245 b/tests/qemu-iotests/245
index 7891a21..d11e73c 100644
--- a/tests/qemu-iotests/245
+++ b/tests/qemu-iotests/245
@@ -859,9 +859,9 @@ class TestBlockdevReopen(iotests.QMPTestCase):
                              device = 'hd0', base_node = 'hd2', speed = 512 * 1024)
         self.assert_qmp(result, 'return', {})
 
-        # We can't remove hd2 while the stream job is ongoing
+        # We can remove hd2 while the stream job is ongoing
         opts['backing']['backing'] = None
-        self.reopen(opts, {}, "Cannot change 'backing' link from 'hd1' to 'hd2'")
+        self.reopen(opts, {})
 
         # We can't remove hd1 while the stream job is ongoing
         opts['backing'] = None
-- 
1.8.3.1

[Qemu-devel] [PATCH v4 3/3] block/stream: introduce a bottom node

[Andrey Shinkevich]

The bottom node is the intermediate block device that has the base as its
backing image. It is used instead of the base node while a block stream
job is running to avoid dependency on the base that may change due to the
parallel jobs. The change may take place due to a filter node as well that
is inserted between the base and the intermediate bottom node. It occurs
when the base node is the top one for another commit or stream job.
After the introduction of the bottom node, don't freeze its backing child,
that's the base, anymore.

Suggested-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Signed-off-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>
---
 block/stream.c         | 49 +++++++++++++++++++++++++------------------------
 tests/qemu-iotests/245 |  4 ++--
 2 files changed, 27 insertions(+), 26 deletions(-)

diff --git a/block/stream.c b/block/stream.c
index e73a320..c39212e 100644
--- a/block/stream.c
+++ b/block/stream.c
@@ -31,7 +31,7 @@ enum {
 
 typedef struct StreamBlockJob {
     BlockJob common;
-    BlockDriverState *base;
+    BlockDriverState *bottom;
     BlockdevOnError on_error;
     char *backing_file_str;
     bool bs_read_only;
@@ -56,7 +56,7 @@ static void stream_abort(Job *job)
 
     if (s->chain_frozen) {
         BlockJob *bjob = &s->common;
-        bdrv_unfreeze_backing_chain(blk_bs(bjob->blk), s->base);
+        bdrv_unfreeze_backing_chain(blk_bs(bjob->blk), s->bottom);
     }
 }
 
@@ -65,11 +65,11 @@ static int stream_prepare(Job *job)
     StreamBlockJob *s = container_of(job, StreamBlockJob, common.job);
     BlockJob *bjob = &s->common;
     BlockDriverState *bs = blk_bs(bjob->blk);
-    BlockDriverState *base = s->base;
+    BlockDriverState *base = backing_bs(s->bottom);
     Error *local_err = NULL;
     int ret = 0;
 
-    bdrv_unfreeze_backing_chain(bs, base);
+    bdrv_unfreeze_backing_chain(bs, s->bottom);
     s->chain_frozen = false;
 
     if (bs->backing) {
@@ -112,7 +112,7 @@ static int coroutine_fn stream_run(Job *job, Error **errp)
     StreamBlockJob *s = container_of(job, StreamBlockJob, common.job);
     BlockBackend *blk = s->common.blk;
     BlockDriverState *bs = blk_bs(blk);
-    BlockDriverState *base = s->base;
+    bool enable_cor = !backing_bs(s->bottom);
     int64_t len;
     int64_t offset = 0;
     uint64_t delay_ns = 0;
@@ -121,7 +121,8 @@ static int coroutine_fn stream_run(Job *job, Error **errp)
     int64_t n = 0; /* bytes */
     void *buf;
 
-    if (!bs->backing) {
+    if (bs == s->bottom) {
+        /* Nothing to stream */
         return 0;
     }
 
@@ -138,7 +139,7 @@ static int coroutine_fn stream_run(Job *job, Error **errp)
      * backing chain since the copy-on-read operation does not take base into
      * account.
      */
-    if (!base) {
+    if (enable_cor) {
         bdrv_enable_copy_on_read(bs);
     }
 
@@ -161,9 +162,8 @@ static int coroutine_fn stream_run(Job *job, Error **errp)
         } else if (ret >= 0) {
             /* Copy if allocated in the intermediate images.  Limit to the
              * known-unallocated area [offset, offset+n*BDRV_SECTOR_SIZE).  */
-            ret = bdrv_is_allocated_above(backing_bs(bs), base, false,
+            ret = bdrv_is_allocated_above(backing_bs(bs), s->bottom, true,
                                           offset, n, &n);
-
             /* Finish early if end of backing file has been reached */
             if (ret == 0 && n == 0) {
                 n = len - offset;
@@ -200,7 +200,7 @@ static int coroutine_fn stream_run(Job *job, Error **errp)
         }
     }
 
-    if (!base) {
+    if (enable_cor) {
         bdrv_disable_copy_on_read(bs);
     }
 
@@ -232,8 +232,10 @@ void stream_start(const char *job_id, BlockDriverState *bs,
     StreamBlockJob *s;
     BlockDriverState *iter;
     bool bs_read_only;
+    int basic_flags = BLK_PERM_CONSISTENT_READ | BLK_PERM_WRITE_UNCHANGED;
+    BlockDriverState *bottom = bdrv_find_overlay(bs, base);
 
-    if (bdrv_freeze_backing_chain(bs, base, errp) < 0) {
+    if (bdrv_freeze_backing_chain(bs, bottom, errp) < 0) {
         return;
     }
 
@@ -250,26 +252,25 @@ void stream_start(const char *job_id, BlockDriverState *bs,
      * already have our own plans. Also don't allow resize as the image size is
      * queried only at the job start and then cached. */
     s = block_job_create(job_id, &stream_job_driver, NULL, bs,
-                         BLK_PERM_CONSISTENT_READ | BLK_PERM_WRITE_UNCHANGED |
-                         BLK_PERM_GRAPH_MOD,
-                         BLK_PERM_CONSISTENT_READ | BLK_PERM_WRITE_UNCHANGED |
-                         BLK_PERM_WRITE,
+                         basic_flags | BLK_PERM_GRAPH_MOD,
+                         basic_flags | BLK_PERM_WRITE,
                          speed, creation_flags, NULL, NULL, errp);
     if (!s) {
         goto fail;
     }
 
-    /* Block all intermediate nodes between bs and base, because they will
-     * disappear from the chain after this operation. The streaming job reads
-     * every block only once, assuming that it doesn't change, so block writes
-     * and resizes. */
-    for (iter = backing_bs(bs); iter && iter != base; iter = backing_bs(iter)) {
-        block_job_add_bdrv(&s->common, "intermediate node", iter, 0,
-                           BLK_PERM_CONSISTENT_READ | BLK_PERM_WRITE_UNCHANGED,
-                           &error_abort);
+    /*
+     * Block all intermediate nodes between bs and bottom (inclusive), because
+     * they will disappear from the chain after this operation. The streaming
+     * job reads every block only once, assuming that it doesn't change, so
+     * forbid writes and resizes.
+     */
+    for (iter = bs; iter != bottom; iter = backing_bs(iter)) {
+        block_job_add_bdrv(&s->common, "intermediate node", backing_bs(iter),
+                           0, basic_flags, &error_abort);
     }
 
-    s->base = base;
+    s->bottom = bottom;
     s->backing_file_str = g_strdup(backing_file_str);
     s->bs_read_only = bs_read_only;
     s->chain_frozen = true;
diff --git a/tests/qemu-iotests/245 b/tests/qemu-iotests/245
index 7891a21..d11e73c 100644
--- a/tests/qemu-iotests/245
+++ b/tests/qemu-iotests/245
@@ -859,9 +859,9 @@ class TestBlockdevReopen(iotests.QMPTestCase):
                              device = 'hd0', base_node = 'hd2', speed = 512 * 1024)
         self.assert_qmp(result, 'return', {})
 
-        # We can't remove hd2 while the stream job is ongoing
+        # We can remove hd2 while the stream job is ongoing
         opts['backing']['backing'] = None
-        self.reopen(opts, {}, "Cannot change 'backing' link from 'hd1' to 'hd2'")
+        self.reopen(opts, {})
 
         # We can't remove hd1 while the stream job is ongoing
         opts['backing'] = None
-- 
1.8.3.1

Re: [Qemu-devel] [PATCH v4 1/3] block: include base when checking image chain for block allocation

[Vladimir Sementsov-Ogievskiy]

08.04.2019 21:22, Andrey Shinkevich wrote:
> This patch is used in the 'block/stream: introduce a bottom node'
> that is following. Instead of the base node, the caller may pass
> the node that has the base as its backing image to the function
> bdrv_is_allocated_above() with a new parameter include_base = true
> and get rid of the dependency on the base that may change during
> commit/stream parallel jobs. Now, if the specified base is not
> found in the backing image chain, the QEMU will crash.
> 
> Suggested-by: Vladimir Sementsov-Ogievskiy<vsementsov@virtuozzo.com>
> Signed-off-by: Andrey Shinkevich<andrey.shinkevich@virtuozzo.com>

Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>

-- 
Best regards,
Vladimir

Re: [Qemu-devel] [PATCH v4 1/3] block: include base when checking image chain for block allocation

[Vladimir Sementsov-Ogievskiy]

08.04.2019 21:22, Andrey Shinkevich wrote:
> This patch is used in the 'block/stream: introduce a bottom node'
> that is following. Instead of the base node, the caller may pass
> the node that has the base as its backing image to the function
> bdrv_is_allocated_above() with a new parameter include_base = true
> and get rid of the dependency on the base that may change during
> commit/stream parallel jobs. Now, if the specified base is not
> found in the backing image chain, the QEMU will crash.
> 
> Suggested-by: Vladimir Sementsov-Ogievskiy<vsementsov@virtuozzo.com>
> Signed-off-by: Andrey Shinkevich<andrey.shinkevich@virtuozzo.com>

Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>

-- 
Best regards,
Vladimir

Re: [Qemu-devel] [PATCH v4 3/3] block/stream: introduce a bottom node

[Vladimir Sementsov-Ogievskiy]

08.04.2019 21:22, Andrey Shinkevich wrote:
> The bottom node is the intermediate block device that has the base as its
> backing image. It is used instead of the base node while a block stream
> job is running to avoid dependency on the base that may change due to the
> parallel jobs. The change may take place due to a filter node as well that
> is inserted between the base and the intermediate bottom node. It occurs
> when the base node is the top one for another commit or stream job.
> After the introduction of the bottom node, don't freeze its backing child,
> that's the base, anymore.
> 
> Suggested-by: Vladimir Sementsov-Ogievskiy<vsementsov@virtuozzo.com>
> Signed-off-by: Andrey Shinkevich<andrey.shinkevich@virtuozzo.com>

Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>

-- 
Best regards,
Vladimir

Re: [Qemu-devel] [PATCH v4 3/3] block/stream: introduce a bottom node

[Vladimir Sementsov-Ogievskiy]

08.04.2019 21:22, Andrey Shinkevich wrote:
> The bottom node is the intermediate block device that has the base as its
> backing image. It is used instead of the base node while a block stream
> job is running to avoid dependency on the base that may change due to the
> parallel jobs. The change may take place due to a filter node as well that
> is inserted between the base and the intermediate bottom node. It occurs
> when the base node is the top one for another commit or stream job.
> After the introduction of the bottom node, don't freeze its backing child,
> that's the base, anymore.
> 
> Suggested-by: Vladimir Sementsov-Ogievskiy<vsementsov@virtuozzo.com>
> Signed-off-by: Andrey Shinkevich<andrey.shinkevich@virtuozzo.com>

Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>

-- 
Best regards,
Vladimir

Re: [Qemu-devel] [PATCH v4 3/3] block/stream: introduce a bottom node

[Alberto Garcia]

On Mon 08 Apr 2019 08:22:21 PM CEST, Andrey Shinkevich wrote:
> The bottom node is the intermediate block device that has the base as its
> backing image. It is used instead of the base node while a block stream
> job is running to avoid dependency on the base that may change due to the
> parallel jobs. The change may take place due to a filter node as well that
> is inserted between the base and the intermediate bottom node. It occurs
> when the base node is the top one for another commit or stream job.
> After the introduction of the bottom node, don't freeze its backing child,
> that's the base, anymore.
>
> Suggested-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
> Signed-off-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>

Reviewed-by: Alberto Garcia <berto@igalia.com>

Berto

Re: [Qemu-devel] [PATCH v4 3/3] block/stream: introduce a bottom node

[Alberto Garcia]

On Mon 08 Apr 2019 08:22:21 PM CEST, Andrey Shinkevich wrote:
> The bottom node is the intermediate block device that has the base as its
> backing image. It is used instead of the base node while a block stream
> job is running to avoid dependency on the base that may change due to the
> parallel jobs. The change may take place due to a filter node as well that
> is inserted between the base and the intermediate bottom node. It occurs
> when the base node is the top one for another commit or stream job.
> After the introduction of the bottom node, don't freeze its backing child,
> that's the base, anymore.
>
> Suggested-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
> Signed-off-by: Andrey Shinkevich <andrey.shinkevich@virtuozzo.com>

Reviewed-by: Alberto Garcia <berto@igalia.com>

Berto

Re: [Qemu-devel] [PATCH v4 1/3] block: include base when checking image chain for block allocation

[Alberto Garcia]

On Mon 08 Apr 2019 08:22:19 PM CEST, Andrey Shinkevich wrote:
>   * Return true if (a prefix of) the given range is allocated in any image
> - * between BASE and TOP (inclusive).  BASE can be NULL to check if the given
> + * between BASE and TOP (TOP included). To check the BASE image, set the
> + * 'include_base' to 'true'. The BASE can be NULL to check if the given
>   * offset is allocated in any image of the chain.  Return false otherwise,
>   * or negative errno on failure.

I'm not a native speaker but that sounds a bit odd to me.

Alternative: 

  * Return true if (a prefix of) the given range is allocated in any image
  * between BASE and TOP (BASE is only included if include_base is set).
  * BASE can be NULL to check if the given offset is allocated in any
  * image of the chain.  Return false otherwise, or negative errno on
  * failure.

> -    while (intermediate && intermediate != base) {
> +    while (include_base || intermediate != base) {
>          int64_t pnum_inter;
>          int64_t size_inter;
>  
> @@ -2360,6 +2364,10 @@ int bdrv_is_allocated_above(BlockDriverState *top,
>              n = pnum_inter;
>          }
>  
> +        if (intermediate == base) {
> +            break;
> +        }
> +
>          intermediate = backing_bs(intermediate);

I find that the new condition + the break make things a bit less
readable. I think it would be simpler with something like this:

   BlockDriverState *stop_at = include_base ? backing_bs(base) : base;

   while (intermediate != stop_at) {
      ...
   }

(stop_at is a terrible name, but I can't think of anything better at the
moment)

Berto

Re: [Qemu-devel] [PATCH v4 1/3] block: include base when checking image chain for block allocation

[Alberto Garcia]

On Mon 08 Apr 2019 08:22:19 PM CEST, Andrey Shinkevich wrote:
>   * Return true if (a prefix of) the given range is allocated in any image
> - * between BASE and TOP (inclusive).  BASE can be NULL to check if the given
> + * between BASE and TOP (TOP included). To check the BASE image, set the
> + * 'include_base' to 'true'. The BASE can be NULL to check if the given
>   * offset is allocated in any image of the chain.  Return false otherwise,
>   * or negative errno on failure.

I'm not a native speaker but that sounds a bit odd to me.

Alternative: 

  * Return true if (a prefix of) the given range is allocated in any image
  * between BASE and TOP (BASE is only included if include_base is set).
  * BASE can be NULL to check if the given offset is allocated in any
  * image of the chain.  Return false otherwise, or negative errno on
  * failure.

> -    while (intermediate && intermediate != base) {
> +    while (include_base || intermediate != base) {
>          int64_t pnum_inter;
>          int64_t size_inter;
>  
> @@ -2360,6 +2364,10 @@ int bdrv_is_allocated_above(BlockDriverState *top,
>              n = pnum_inter;
>          }
>  
> +        if (intermediate == base) {
> +            break;
> +        }
> +
>          intermediate = backing_bs(intermediate);

I find that the new condition + the break make things a bit less
readable. I think it would be simpler with something like this:

   BlockDriverState *stop_at = include_base ? backing_bs(base) : base;

   while (intermediate != stop_at) {
      ...
   }

(stop_at is a terrible name, but I can't think of anything better at the
moment)

Berto

Re: [Qemu-devel] [PATCH v4 1/3] block: include base when checking image chain for block allocation

[Vladimir Sementsov-Ogievskiy]

09.04.2019 17:18, Alberto Garcia wrote:
> On Mon 08 Apr 2019 08:22:19 PM CEST, Andrey Shinkevich wrote:
>>    * Return true if (a prefix of) the given range is allocated in any image
>> - * between BASE and TOP (inclusive).  BASE can be NULL to check if the given
>> + * between BASE and TOP (TOP included). To check the BASE image, set the
>> + * 'include_base' to 'true'. The BASE can be NULL to check if the given
>>    * offset is allocated in any image of the chain.  Return false otherwise,
>>    * or negative errno on failure.
> 
> I'm not a native speaker but that sounds a bit odd to me.
> 
> Alternative:
> 
>    * Return true if (a prefix of) the given range is allocated in any image
>    * between BASE and TOP (BASE is only included if include_base is set).
>    * BASE can be NULL to check if the given offset is allocated in any
>    * image of the chain.  Return false otherwise, or negative errno on
>    * failure.
> 
>> -    while (intermediate && intermediate != base) {
>> +    while (include_base || intermediate != base) {
>>           int64_t pnum_inter;
>>           int64_t size_inter;
>>   
>> @@ -2360,6 +2364,10 @@ int bdrv_is_allocated_above(BlockDriverState *top,
>>               n = pnum_inter;
>>           }
>>   
>> +        if (intermediate == base) {
>> +            break;
>> +        }
>> +
>>           intermediate = backing_bs(intermediate);
> 
> I find that the new condition + the break make things a bit less
> readable. I think it would be simpler with something like this:
> 
>     BlockDriverState *stop_at = include_base ? backing_bs(base) : base;
> 
>     while (intermediate != stop_at) {
>        ...
>     }
> 

But in this way you return back dependence on base, which we don't freeze and which
may disappear on some iteration. We should not touch backing_bs(base) in any way.


-- 
Best regards,
Vladimir

Re: [Qemu-devel] [PATCH v4 1/3] block: include base when checking image chain for block allocation

[Vladimir Sementsov-Ogievskiy]

09.04.2019 17:18, Alberto Garcia wrote:
> On Mon 08 Apr 2019 08:22:19 PM CEST, Andrey Shinkevich wrote:
>>    * Return true if (a prefix of) the given range is allocated in any image
>> - * between BASE and TOP (inclusive).  BASE can be NULL to check if the given
>> + * between BASE and TOP (TOP included). To check the BASE image, set the
>> + * 'include_base' to 'true'. The BASE can be NULL to check if the given
>>    * offset is allocated in any image of the chain.  Return false otherwise,
>>    * or negative errno on failure.
> 
> I'm not a native speaker but that sounds a bit odd to me.
> 
> Alternative:
> 
>    * Return true if (a prefix of) the given range is allocated in any image
>    * between BASE and TOP (BASE is only included if include_base is set).
>    * BASE can be NULL to check if the given offset is allocated in any
>    * image of the chain.  Return false otherwise, or negative errno on
>    * failure.
> 
>> -    while (intermediate && intermediate != base) {
>> +    while (include_base || intermediate != base) {
>>           int64_t pnum_inter;
>>           int64_t size_inter;
>>   
>> @@ -2360,6 +2364,10 @@ int bdrv_is_allocated_above(BlockDriverState *top,
>>               n = pnum_inter;
>>           }
>>   
>> +        if (intermediate == base) {
>> +            break;
>> +        }
>> +
>>           intermediate = backing_bs(intermediate);
> 
> I find that the new condition + the break make things a bit less
> readable. I think it would be simpler with something like this:
> 
>     BlockDriverState *stop_at = include_base ? backing_bs(base) : base;
> 
>     while (intermediate != stop_at) {
>        ...
>     }
> 

But in this way you return back dependence on base, which we don't freeze and which
may disappear on some iteration. We should not touch backing_bs(base) in any way.


-- 
Best regards,
Vladimir

Re: [Qemu-devel] [PATCH v4 1/3] block: include base when checking image chain for block allocation

[Alberto Garcia]

On Tue 09 Apr 2019 04:43:12 PM CEST, Vladimir Sementsov-Ogievskiy wrote:
>>> -    while (intermediate && intermediate != base) {
>>> +    while (include_base || intermediate != base) {
>>>           int64_t pnum_inter;
>>>           int64_t size_inter;
>>>   
>>> @@ -2360,6 +2364,10 @@ int bdrv_is_allocated_above(BlockDriverState *top,
>>>               n = pnum_inter;
>>>           }
>>>   
>>> +        if (intermediate == base) {
>>> +            break;
>>> +        }
>>> +
>>>           intermediate = backing_bs(intermediate);
>> 
>> I find that the new condition + the break make things a bit less
>> readable. I think it would be simpler with something like this:
>> 
>>     BlockDriverState *stop_at = include_base ? backing_bs(base) : base;
>> 
>>     while (intermediate != stop_at) {
>>        ...
>>     }
>> 
>
> But in this way you return back dependence on base, which we don't
> freeze and which may disappear on some iteration. We should not touch
> backing_bs(base) in any way.

Ok, I see.

Reviewed-by: Alberto Garcia <berto@igalia.com>

(feel free to edit the comment with my suggestion, or leave it as it is
if you prefer)

Berto

Re: [Qemu-devel] [PATCH v4 1/3] block: include base when checking image chain for block allocation

[Alberto Garcia]

On Tue 09 Apr 2019 04:43:12 PM CEST, Vladimir Sementsov-Ogievskiy wrote:
>>> -    while (intermediate && intermediate != base) {
>>> +    while (include_base || intermediate != base) {
>>>           int64_t pnum_inter;
>>>           int64_t size_inter;
>>>   
>>> @@ -2360,6 +2364,10 @@ int bdrv_is_allocated_above(BlockDriverState *top,
>>>               n = pnum_inter;
>>>           }
>>>   
>>> +        if (intermediate == base) {
>>> +            break;
>>> +        }
>>> +
>>>           intermediate = backing_bs(intermediate);
>> 
>> I find that the new condition + the break make things a bit less
>> readable. I think it would be simpler with something like this:
>> 
>>     BlockDriverState *stop_at = include_base ? backing_bs(base) : base;
>> 
>>     while (intermediate != stop_at) {
>>        ...
>>     }
>> 
>
> But in this way you return back dependence on base, which we don't
> freeze and which may disappear on some iteration. We should not touch
> backing_bs(base) in any way.

Ok, I see.

Reviewed-by: Alberto Garcia <berto@igalia.com>

(feel free to edit the comment with my suggestion, or leave it as it is
if you prefer)

Berto