aio_co_wake() is generally safe to call regardless of whether the
coroutine is already yielding or not.  If it is not yet yielding, it
will be scheduled to run when it does yield.

Caveats:
- The caller must be independent of the coroutine (to ensure the
  coroutine must be yielding if both are in the same AioContext), i.e.
  must not be the same coroutine
- The coroutine must yield at some point

Make note of this so callers can reason that their use is safe.

Signed-off-by: Hanna Czenczek <hreitz@redhat.com>
---
 include/block/aio.h | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/include/block/aio.h b/include/block/aio.h
index 99ff48420b..6ed97f0a4c 100644
--- a/include/block/aio.h
+++ b/include/block/aio.h
@@ -650,6 +650,21 @@ void coroutine_fn aio_co_reschedule_self(AioContext *new_ctx);
  * aio_co_wake may be executed either in coroutine or non-coroutine
  * context.  The coroutine must not be entered by anyone else while
  * aio_co_wake() is active.
+ *
+ * If `co`'s AioContext differs from the current AioContext, this will call
+ * aio_co_schedule(), which makes this safe to use even when `co` has not
+ * yielded yet.  In such a case, it will be entered once it yields.
+ *
+ * In contrast, if `co`'s AioContext is equal to the current one, it is
+ * required for `co` to currently be yielding.  This is generally the case
+ * if the caller is not in `co` (i.e. invoked by `co`), because the only
+ * other way for the caller to be running then is for `co` to currently be
+ * yielding.
+ *
+ * Therefore, if there is no way for the caller to be invoked/entered by
+ * `co`, it is generally safe to call this regardless of whether `co` is
+ * known to already be yielding or not -- it only has to yield at some
+ * point.
  */
 void aio_co_wake(Coroutine *co);
 
-- 
2.51.1