[PATCH] block: remove unused BLOCK_OP_TYPE_DATAPLANE

[PATCH] block: remove unused BLOCK_OP_TYPE_DATAPLANE

[Stefan Hajnoczi]

BLOCK_OP_TYPE_DATAPLANE prevents BlockDriverState from being used by
virtio-blk/virtio-scsi with IOThread. Commit b112a65c52aa ("block:
declare blockjobs and dataplane friends!") eliminated the main reason
for this blocker in 2014.

Nowadays the block layer supports I/O from multiple AioContexts, so
there is even less reason to block IOThread users. Any legitimate
reasons related to interference would probably also apply to
non-IOThread users.

The only remaining users are bdrv_op_unblock(BLOCK_OP_TYPE_DATAPLANE)
calls after bdrv_op_block_all(). If we remove BLOCK_OP_TYPE_DATAPLANE
their behavior doesn't change.

Existing bdrv_op_block_all() callers that don't explicitly unblock
BLOCK_OP_TYPE_DATAPLANE seem to do so simply because no one bothered to
rather than because it is necessary to keep BLOCK_OP_TYPE_DATAPLANE
blocked.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 include/block/block-common.h | 1 -
 block/replication.c          | 1 -
 blockjob.c                   | 2 --
 hw/block/virtio-blk.c        | 9 ---------
 hw/scsi/virtio-scsi.c        | 3 ---
 5 files changed, 16 deletions(-)

diff --git a/include/block/block-common.h b/include/block/block-common.h
index 338fe5ff7a..2b0aae5bc9 100644
--- a/include/block/block-common.h
+++ b/include/block/block-common.h
@@ -355,7 +355,6 @@ typedef enum BlockOpType {
     BLOCK_OP_TYPE_CHANGE,
     BLOCK_OP_TYPE_COMMIT_SOURCE,
     BLOCK_OP_TYPE_COMMIT_TARGET,
-    BLOCK_OP_TYPE_DATAPLANE,
     BLOCK_OP_TYPE_DRIVE_DEL,
     BLOCK_OP_TYPE_EJECT,
     BLOCK_OP_TYPE_EXTERNAL_SNAPSHOT,
diff --git a/block/replication.c b/block/replication.c
index 2ce16f0589..d4d677a902 100644
--- a/block/replication.c
+++ b/block/replication.c
@@ -576,7 +576,6 @@ static void replication_start(ReplicationState *rs, ReplicationMode mode,
             return;
         }
         bdrv_op_block_all(top_bs, s->blocker);
-        bdrv_op_unblock(top_bs, BLOCK_OP_TYPE_DATAPLANE, s->blocker);
 
         bdrv_graph_wrunlock();
 
diff --git a/blockjob.c b/blockjob.c
index e94a840d7f..32007f31a9 100644
--- a/blockjob.c
+++ b/blockjob.c
@@ -539,8 +539,6 @@ void *block_job_create(const char *job_id, const BlockJobDriver *driver,
         goto fail;
     }
 
-    bdrv_op_unblock(bs, BLOCK_OP_TYPE_DATAPLANE, job->blocker);
-
     if (!block_job_set_speed(job, speed, errp)) {
         goto fail;
     }
diff --git a/hw/block/virtio-blk.c b/hw/block/virtio-blk.c
index e0acce89e1..a1829e3abd 100644
--- a/hw/block/virtio-blk.c
+++ b/hw/block/virtio-blk.c
@@ -1562,15 +1562,6 @@ static bool virtio_blk_vq_aio_context_init(VirtIOBlock *s, Error **errp)
             error_setg(errp, "ioeventfd is required for iothread");
             return false;
         }
-
-        /*
-         * If ioeventfd is (re-)enabled while the guest is running there could
-         * be block jobs that can conflict.
-         */
-        if (blk_op_is_blocked(conf->conf.blk, BLOCK_OP_TYPE_DATAPLANE, errp)) {
-            error_prepend(errp, "cannot start virtio-blk ioeventfd: ");
-            return false;
-        }
     }
 
     s->vq_aio_context = g_new(AioContext *, conf->num_queues);
diff --git a/hw/scsi/virtio-scsi.c b/hw/scsi/virtio-scsi.c
index 23516995dc..7d094e1881 100644
--- a/hw/scsi/virtio-scsi.c
+++ b/hw/scsi/virtio-scsi.c
@@ -1065,9 +1065,6 @@ static void virtio_scsi_hotplug(HotplugHandler *hotplug_dev, DeviceState *dev,
     int ret;
 
     if (s->ctx && !s->dataplane_fenced) {
-        if (blk_op_is_blocked(sd->conf.blk, BLOCK_OP_TYPE_DATAPLANE, errp)) {
-            return;
-        }
         ret = blk_set_aio_context(sd->conf.blk, s->ctx, errp);
         if (ret < 0) {
             return;
-- 
2.48.1

Re: [PATCH] block: remove unused BLOCK_OP_TYPE_DATAPLANE

[Eric Blake]

On Mon, Feb 03, 2025 at 01:25:29PM -0500, Stefan Hajnoczi wrote:
> BLOCK_OP_TYPE_DATAPLANE prevents BlockDriverState from being used by
> virtio-blk/virtio-scsi with IOThread. Commit b112a65c52aa ("block:
> declare blockjobs and dataplane friends!") eliminated the main reason
> for this blocker in 2014.

Wow, that's a long time.

> 
> Nowadays the block layer supports I/O from multiple AioContexts, so
> there is even less reason to block IOThread users. Any legitimate
> reasons related to interference would probably also apply to
> non-IOThread users.
> 
> The only remaining users are bdrv_op_unblock(BLOCK_OP_TYPE_DATAPLANE)
> calls after bdrv_op_block_all(). If we remove BLOCK_OP_TYPE_DATAPLANE
> their behavior doesn't change.
> 
> Existing bdrv_op_block_all() callers that don't explicitly unblock
> BLOCK_OP_TYPE_DATAPLANE seem to do so simply because no one bothered to
> rather than because it is necessary to keep BLOCK_OP_TYPE_DATAPLANE
> blocked.
> 
> Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
> ---
>  include/block/block-common.h | 1 -
>  block/replication.c          | 1 -
>  blockjob.c                   | 2 --
>  hw/block/virtio-blk.c        | 9 ---------
>  hw/scsi/virtio-scsi.c        | 3 ---
>  5 files changed, 16 deletions(-)

Reviewed-by: Eric Blake <eblake@redhat.com>

-- 
Eric Blake, Principal Software Engineer
Red Hat, Inc.
Virtualization:  qemu.org | libguestfs.org

Re: [PATCH] block: remove unused BLOCK_OP_TYPE_DATAPLANE

[Kevin Wolf]

Am 03.02.2025 um 19:25 hat Stefan Hajnoczi geschrieben:
> BLOCK_OP_TYPE_DATAPLANE prevents BlockDriverState from being used by
> virtio-blk/virtio-scsi with IOThread. Commit b112a65c52aa ("block:
> declare blockjobs and dataplane friends!") eliminated the main reason
> for this blocker in 2014.
> 
> Nowadays the block layer supports I/O from multiple AioContexts, so
> there is even less reason to block IOThread users. Any legitimate
> reasons related to interference would probably also apply to
> non-IOThread users.
> 
> The only remaining users are bdrv_op_unblock(BLOCK_OP_TYPE_DATAPLANE)
> calls after bdrv_op_block_all(). If we remove BLOCK_OP_TYPE_DATAPLANE
> their behavior doesn't change.
> 
> Existing bdrv_op_block_all() callers that don't explicitly unblock
> BLOCK_OP_TYPE_DATAPLANE seem to do so simply because no one bothered to
> rather than because it is necessary to keep BLOCK_OP_TYPE_DATAPLANE
> blocked.
> 
> Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

Thanks, applied to the block branch.

I've actually had pretty much the same patch lying around for.. *checks*
five years, but never sent it because I intended it to be part of a more
general op blocker removal series. I don't think we actually rely on any
of them any more, but proving it was still hard when I tried.

One additional part my patch had is removing blk_op_is_blocked() because
it's now unused. I'll send this as a separate patch.

Kevin