Am 10.04.2017 um 10:45 hat Fam Zheng geschrieben:
> On Mon, 04/10 10:06, Kevin Wolf wrote:
> > Am 08.04.2017 um 05:43 hat Fam Zheng geschrieben:
> > > On Fri, 04/07 13:50, Stefan Hajnoczi wrote:
> > > > On Fri, Apr 07, 2017 at 02:54:12PM +0800, Fam Zheng wrote:
> > > > > @@ -4413,6 +4416,10 @@ void bdrv_set_aio_context(BlockDriverState
> > > > > *bs, AioContext *new_context)
> > > > > aio_context_acquire(new_context);
> > > > > bdrv_attach_aio_context(bs, new_context);
> > > > > aio_context_release(new_context);
> > > > > + if (bs->job) {
> > > > > + block_job_resume(bs->job);
> > > > > + }
> > > >
> > > > Should this be called before aio_context_release(new_context)?
> > >
> > > Yes, and I'm going to replace it with bdrv_parent_drained_begin() as Kevin
> > > suggested.
> >
> > I think at the moment bdrv_parent_drained_begin() can't replace it yet,
> > but you need both.
>
> I think we have it already, see 600ac6a0e (blockjob: add devops to blockjob
> backends):
>
> bdrv_parent_drained_begin
> -> blk_root_drained_begin
> -> block_job_drained_begin
> -> block_job_pause
Ah, yes, you're right. Somehow I thought this was only for 2.10.
Kevin
