On Fri, Jul 10, 2015 at 08:46:44AM +0800, Fam Zheng wrote:
> On Wed, 07/08 11:58, Stefan Hajnoczi wrote:
> > On Wed, Jul 08, 2015 at 09:01:27AM +0800, Fam Zheng wrote:
> > > On Tue, 07/07 16:08, Stefan Hajnoczi wrote:
> > > > > +#define EPOLL_BATCH 128
> > > > > +static bool aio_poll_epoll(AioConte
On Wed, 07/08 11:58, Stefan Hajnoczi wrote:
> On Wed, Jul 08, 2015 at 09:01:27AM +0800, Fam Zheng wrote:
> > On Tue, 07/07 16:08, Stefan Hajnoczi wrote:
> > > > +#define EPOLL_BATCH 128
> > > > +static bool aio_poll_epoll(AioContext *ctx, bool blocking)
> > > > +{
> > > > +AioHandler *node;
> >
On Wed, Jul 08, 2015 at 09:01:27AM +0800, Fam Zheng wrote:
> On Tue, 07/07 16:08, Stefan Hajnoczi wrote:
> > > +#define EPOLL_BATCH 128
> > > +static bool aio_poll_epoll(AioContext *ctx, bool blocking)
> > > +{
> > > +AioHandler *node;
> > > +bool was_dispatching;
> > > +int i, ret;
> >
On Tue, 07/07 16:08, Stefan Hajnoczi wrote:
> > @@ -44,6 +47,12 @@ static AioHandler *find_aio_handler(AioContext *ctx, int
> > fd)
> >
> > void aio_context_setup(AioContext *ctx, Error **errp)
> > {
> > +#ifdef CONFIG_EPOLL
> > +ctx->epollfd = epoll_create1(EPOLL_CLOEXEC);
> > +if (ct
On 07/07/2015 17:08, Stefan Hajnoczi wrote:
>> > +
>> > +#define EPOLL_BATCH 128
>> > +static bool aio_poll_epoll(AioContext *ctx, bool blocking)
>> > +{
>> > +AioHandler *node;
>> > +bool was_dispatching;
>> > +int i, ret;
>> > +bool progress;
>> > +int64_t timeout;
>> > +
On Tue, Jun 30, 2015 at 09:19:45PM +0800, Fam Zheng wrote:
> =
> # of scsi-disks |master | epoll
>| rd wrrandrw | rdwrrandrw
> ---
This patch let aio_poll use epoll_wait(2) syscall instead of
qemu_poll_ns, if possible. It improves scalability of
iothread (for example, virtio-scsi-dataplane.)
The epollfd is managed together with the GSource and ctx->aio_handlers,
by creating epoll_event instances for each watched aio fd and ad
