Re: [Qemu-devel] [PATCH] block/throttle: Use host clock type

[Fam Zheng <famz@redhat.com>]

On Fri, 03/13 13:28, Paolo Bonzini wrote:
> 
> 
> On 13/03/2015 13:23, Alberto Garcia wrote:
> > On Fri, Mar 13, 2015 at 02:35:29PM +0800, Fam Zheng wrote:
> > 
> >> Throttle timers won't make any progress when VCPU is not running,
> >> which is prone to stall the request queue in cases like utils,
> >> qtest, suspending, and live migration, unless carefully handled.
> > 
> > Yes, this can be easily reproduced by stopping the VM and starting a
> > block-commit job. If the I/O in that device is throttled then the job
> > will be stalled.
> 
> That may be a different bug.  Should jobs be subject to throttling at all?

You are asking about the next bug below, right?

I lean towards yes, but the problem is that we are in the middle of yes and no.
Before "stop" it is throttled, but upon "stop", it is drained right to the end,
unthrottled.

Fam

> 
> Paolo
> 
> > Then there's also the situation that we discussed in IRC: if the
> > block-commit job is ongoing and then we stop the VM, then the rest of
> > the data will be committed bypassing the throttling settings.  But
> > that's not related to these changes.
> > 
> >> Signed-off-by: Fam Zheng <famz@redhat.com>
> >> ---
> >>  block.c               |  2 +-
> >>  tests/test-throttle.c | 14 +++++++-------
> >>  2 files changed, 8 insertions(+), 8 deletions(-)
> > 
> > Reviewed-By: Alberto Garcia <berto@igalia.com>
> > 
> > Berto
> > 
> > 
>