Am 13.09.2018 um 17:15 hat Paolo Bonzini geschrieben: > On 13/09/2018 14:52, Kevin Wolf wrote: > > Also, job_drain() without aio_poll() isn't necessarily enough to make > > progress on a job, it could depend on bottom halves to be executed. > > This is not changed by the patch though; AIO_WAIT_WHILE does not do the > bottom halves part anymore, bdrv_drain_poll_top_level does it instead. > > What is it that needs bottom halves to be executed? Could it delay > switching the job state until after that?
AIO_WAIT_WHILE() does still execute bottom halves if the condition is true, which is all that is needed to make progress. It only doesn't execute them unconditionally any more. Kevin
