On 10/19/2015 07:21 AM, Pedro Alves wrote:
> On 10/16/2015 08:30 PM, Luis Machado wrote:
>
>> How does the attached update look?
>
> Is there a reason to keep the "rs->waiting_for_stop_reply = 1;" lines?

No, that was a mistake since we inverted the logic. What about the 
following?

Now we'll only set rs->waiting_for_stop_reply to 1 when resuming in 
"all-stop" mode.