block: add support for job pause/resume

Job pausing reuses the existing support for cancellable sleeps.  A pause
happens at the next sleeping point and lasts until the coroutine is
re-entered explicitly.  Cancellation was already doing a forced resume,
so implement it explicitly in terms of resume.

Paused jobs cannot be canceled without first resuming them.  This ensures
that I/O errors are never missed by management.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

5 files changed, 72 insertions, 6 deletions

diff --git a/blockdev.c b/blockdev.c
index 9a98ce9c89..612dd71f2d 100644
--- a/blockdev.c
+++ b/blockdev.c
@@ -1221,6 +1221,10 @@ void qmp_block_job_cancel(const char *device, Error **errp)
         error_set(errp, QERR_BLOCK_JOB_NOT_ACTIVE, device);
         return;
     }
+    if (job->paused) {
+        error_set(errp, QERR_BLOCK_JOB_PAUSED, device);
+        return;
+    }
 
     trace_qmp_block_job_cancel(job);
     block_job_cancel(job);
diff --git a/blockjob.c b/blockjob.c
index 64c9d2dda1..8219f73979 100644
--- a/blockjob.c
+++ b/blockjob.c
@@ -99,14 +99,30 @@ void block_job_set_speed(BlockJob *job, int64_t speed, Error **errp)
     job->speed = speed;
 }
 
-void block_job_cancel(BlockJob *job)
+void block_job_pause(BlockJob *job)
 {
-    job->cancelled = true;
+    job->paused = true;
+}
+
+bool block_job_is_paused(BlockJob *job)
+{
+    return job->paused;
+}
+
+void block_job_resume(BlockJob *job)
+{
+    job->paused = false;
     if (job->co && !job->busy) {
         qemu_coroutine_enter(job->co, NULL);
     }
 }
 
+void block_job_cancel(BlockJob *job)
+{
+    job->cancelled = true;
+    block_job_resume(job);
+}
+
 bool block_job_is_cancelled(BlockJob *job)
 {
     return job->cancelled;
@@ -154,12 +170,20 @@ int block_job_cancel_sync(BlockJob *job)
 
 void block_job_sleep_ns(BlockJob *job, QEMUClock *clock, int64_t ns)
 {
+    assert(job->busy);
+
     /* Check cancellation *before* setting busy = false, too!  */
-    if (!block_job_is_cancelled(job)) {
-        job->busy = false;
+    if (block_job_is_cancelled(job)) {
+        return;
+    }
+
+    job->busy = false;
+    if (block_job_is_paused(job)) {
+        qemu_coroutine_yield();
+    } else {
         co_sleep_ns(clock, ns);
-        job->busy = true;
     }
+    job->busy = true;
 }
 
 BlockJobInfo *block_job_query(BlockJob *job)
@@ -169,6 +193,7 @@ BlockJobInfo *block_job_query(BlockJob *job)
     info->device = g_strdup(bdrv_get_device_name(job->bs));
     info->len    = job->len;
     info->busy   = job->busy;
+    info->paused = job->paused;
     info->offset = job->offset;
     info->speed  = job->speed;
     return info;
diff --git a/blockjob.h b/blockjob.h
index f3d8d58ce3..ece5afa75b 100644
--- a/blockjob.h
+++ b/blockjob.h
@@ -70,6 +70,12 @@ struct BlockJob {
     bool cancelled;
 
     /**
+     * Set to true if the job is either paused, or will pause itself
+     * as soon as possible (if busy == true).
+     */
+    bool paused;
+
+    /**
      * Set to false by the job while it is in a quiescent state, where
      * no I/O is pending and the job has yielded on any condition
      * that is not detected by #qemu_aio_wait, such as a timer.
@@ -171,6 +177,31 @@ bool block_job_is_cancelled(BlockJob *job);
 BlockJobInfo *block_job_query(BlockJob *job);
 
 /**
+ * block_job_pause:
+ * @job: The job to be paused.
+ *
+ * Asynchronously pause the specified job.
+ */
+void block_job_pause(BlockJob *job);
+
+/**
+ * block_job_resume:
+ * @job: The job to be resumed.
+ *
+ * Resume the specified job.
+ */
+void block_job_resume(BlockJob *job);
+
+/**
+ * block_job_is_paused:
+ * @job: The job being queried.
+ *
+ * Returns whether the job is currently paused, or will pause
+ * as soon as it reaches a sleeping point.
+ */
+bool block_job_is_paused(BlockJob *job);
+
+/**
  * block_job_cancel_sync:
  * @job: The job to be canceled.
  *
diff --git a/qapi-schema.json b/qapi-schema.json
index 6fc6edaa2d..86a6c7fe93 100644
--- a/qapi-schema.json
+++ b/qapi-schema.json
@@ -1101,6 +1101,9 @@
 # @busy: false if the job is known to be in a quiescent state, with
 #        no pending I/O.  Since 1.3.
 #
+# @paused: whether the job is paused or, if @busy is true, will
+#          pause itself as soon as possible.  Since 1.3.
+#
 # @offset: the current progress value
 #
 # @speed: the rate limit, bytes per second
@@ -1109,7 +1112,7 @@
 ##
 { 'type': 'BlockJobInfo',
   'data': {'type': 'str', 'device': 'str', 'len': 'int',
-           'offset': 'int', 'busy': 'bool', 'speed': 'int'} }
+           'offset': 'int', 'busy': 'bool', 'paused': 'bool', 'speed': 'int'} }
 
 ##
 # @query-block-jobs:
diff --git a/qerror.h b/qerror.h
index 485c773662..c91708cc3c 100644
--- a/qerror.h
+++ b/qerror.h
@@ -51,6 +51,9 @@ void assert_no_error(Error *err);
 #define QERR_BLOCK_JOB_NOT_ACTIVE \
     ERROR_CLASS_DEVICE_NOT_ACTIVE, "No active block job on device '%s'"
 
+#define QERR_BLOCK_JOB_PAUSED \
+    ERROR_CLASS_GENERIC_ERROR, "The block job for device '%s' is currently paused"
+
 #define QERR_BLOCK_FORMAT_FEATURE_NOT_SUPPORTED \
     ERROR_CLASS_GENERIC_ERROR, "Block format '%s' used by device '%s' does not support feature '%s'"