On 2012-02-15 13:53, Luiz Capitulino wrote: > On Wed, 15 Feb 2012 10:02:54 +0100 > Jan Kiszka <jan.kis...@siemens.com> wrote: > >> On 2012-02-10 20:31, Luiz Capitulino wrote: >>> The Monitor object is passed back and forth within the migration/savevm >>> code so that it can print errors and progress to the user. >>> >>> However, that approach assumes a HMP monitor, being completely invalid >>> in QMP. >>> >>> This commit drops almost every single usage of the Monitor object, all >>> monitor_printf() calls have been converted into DPRINTF() ones. >> >> Particularly NACK on this. Either the information is useless anyway, >> then remove it. Otherwise, keep it for channels that can properly >> display it (AKA HMP). I bet the latter can easily be achieved by >> providing non-printing Monitor objects over QMP instances. > > I will consider dropping it :) > > I can think of two ways of displaying status in HMP (considering the new > HMP/QMP split design): > > 1. We add all progress status information to a query command, and let HMP > poll it (manually by the user or automatically from a timer)
As migration can also be synchronous, polling has to be time-driven. > > 2. We emit an event whenever the progress status changes (seems a bit > overkill) If there is a use in QMP as well... dunno. Jan -- Siemens AG, Corporate Technology, CT T DE IT 1 Corporate Competence Center Embedded Linux
