On Mon, Feb 29, 2016 at 01:15:48PM +0800, Jason Wang wrote: > > > On 02/28/2016 10:09 PM, Michael S. Tsirkin wrote: > > On Fri, Feb 26, 2016 at 04:42:44PM +0800, Jason Wang wrote: > >> > This patch tries to poll for new added tx buffer or socket receive > >> > queue for a while at the end of tx/rx processing. The maximum time > >> > spent on polling were specified through a new kind of vring ioctl. > >> > > >> > Signed-off-by: Jason Wang <jasowang@xxxxxxxxxx> > > Looks good overall, but I still see one problem. > > > >> > --- > >> > drivers/vhost/net.c | 79 +++++++++++++++++++++++++++++++++++++++++++--- > >> > drivers/vhost/vhost.c | 14 ++++++++ > >> > drivers/vhost/vhost.h | 1 + > >> > include/uapi/linux/vhost.h | 6 ++++ > >> > 4 files changed, 95 insertions(+), 5 deletions(-) > >> > > >> > diff --git a/drivers/vhost/net.c b/drivers/vhost/net.c > >> > index 9eda69e..c91af93 100644 > >> > --- a/drivers/vhost/net.c > >> > +++ b/drivers/vhost/net.c > >> > @@ -287,6 +287,44 @@ static void vhost_zerocopy_callback(struct ubuf_info *ubuf, bool success) > >> > rcu_read_unlock_bh(); > >> > } > >> > > >> > +static inline unsigned long busy_clock(void) > >> > +{ > >> > + return local_clock() >> 10; > >> > +} > >> > + > >> > +static bool vhost_can_busy_poll(struct vhost_dev *dev, > >> > + unsigned long endtime) > >> > +{ > >> > + return likely(!need_resched()) && > >> > + likely(!time_after(busy_clock(), endtime)) && > >> > + likely(!signal_pending(current)) && > >> > + !vhost_has_work(dev) && > >> > + single_task_running(); > > So I find it quite unfortunate that this still uses single_task_running. > > This means that for example a SCHED_IDLE task will prevent polling from > > becoming active, and that seems like a bug, or at least > > an undocumented feature :). > > Yes, it may need more thoughts. > > > > > Unfortunately this logic affects the behaviour as observed > > by userspace, so we can't merge it like this and tune > > afterwards, since otherwise mangement tools will start > > depending on this logic. > > > > > > How about remove single_task_running() first here and optimize on top? > We probably need something like this to handle overcommitment. Sounds good. -- MST -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html