On 17.07.20 17:24, Vladimir Sementsov-Ogievskiy wrote:
> 17.07.2020 17:00, Max Reitz wrote:
>> On 01.06.20 20:11, Vladimir Sementsov-Ogievskiy wrote:
>>> We'll need async block-copy invocation to use in backup directly.
>>>
>>> Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
>>> ---
>>>   include/block/block-copy.h | 13 +++++++++++++
>>>   block/block-copy.c         | 40 ++++++++++++++++++++++++++++++++++++++
>>>   2 files changed, 53 insertions(+)

[...]

>>> +BlockCopyCallState *block_copy_async(BlockCopyState *s,
>>> +                                     int64_t offset, int64_t bytes,
>>> +                                     bool ratelimit, int max_workers,
>>> +                                     int64_t max_chunk,
>>> +                                     BlockCopyAsyncCallbackFunc cb)
>>> +{
>>> +    BlockCopyCallState *call_state = g_new(BlockCopyCallState, 1);
>>> +    Coroutine *co = qemu_coroutine_create(block_copy_async_co_entry,
>>> +                                          call_state);
>>> +
>>> +    *call_state = (BlockCopyCallState) {
>>> +        .s = s,
>>> +        .offset = offset,
>>> +        .bytes = bytes,
>>> +        .cb = cb,
>>> +    };
>>> +
>>> +    qemu_coroutine_enter(co);
>>
>> Do we need/want any already-in-coroutine shenanigans here?
>>
> 
> No: the aim of the function is to start a new coroutine in parallel,
> independently of are we already in some other coroutine or not.

OK, that makes sense.