OASIS Virtual I/O Device (VIRTIO) TC

Timer/Clock device ------------------ Key: VIRTIO-45 URL: http://tools.oasis-open.org/issues/browse/VIRTIO-45 Project: OASIS Virtual I/O Device (VIRTIO) TC Issue Type: New Feature Reporter: Daniel Kiper There is no timer/clock device in VIRTIO specification. Maybe we should consider addition of one. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://tools.oasis-open.org/issues/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

On Thu, Oct 24, 2013 at 12:47:39PM +0000, OASIS Issues Tracker wrote: > Timer/Clock device > ------------------ > > Key: VIRTIO-45 > URL: http://tools.oasis-open.org/issues/browse/VIRTIO-45 > Project: OASIS Virtual I/O Device (VIRTIO) TC > Issue Type: New Feature > Reporter: Daniel Kiper > > > There is no timer/clock device in VIRTIO specification. Maybe we should consider addition of one. General idea is that if we build hardware independent block, network and other devices then maybe it makes sens to build hardware independent timer/clock. As I know exiting hardware timers/clocks are quite difficult to emulate properly in virtual environments. Most problems are solved right now but maybe we should still consider addition of timer/clock spec in new VIRTIO spec. This way we will make stable time source for virtual guests without issues and limitations imposed by existing and new relevant hardware devices. Daniel

On Tue, Oct 29, 2013 at 01:13:20PM +0100, Daniel Kiper wrote: > On Thu, Oct 24, 2013 at 12:47:39PM +0000, OASIS Issues Tracker wrote: > > Timer/Clock device > > ------------------ > > > > Key: VIRTIO-45 > > URL: http://tools.oasis-open.org/issues/browse/VIRTIO-45 > > Project: OASIS Virtual I/O Device (VIRTIO) TC > > Issue Type: New Feature > > Reporter: Daniel Kiper > > > > > > There is no timer/clock device in VIRTIO specification. Maybe we should consider addition of one. > > General idea is that if we build hardware independent block, network > and other devices then maybe it makes sens to build hardware independent > timer/clock. As I know exiting hardware timers/clocks are quite difficult > to emulate properly in virtual environments. Most problems are solved > right now but maybe we should still consider addition of timer/clock spec > in new VIRTIO spec. This way we will make stable time source for virtual > guests without issues and limitations imposed by existing and new relevant > hardware devices. Do we want that kind of device in VIRTIO spec? Daniel

On Tue, Nov 05, 2013 at 11:42:56AM +0100, Daniel Kiper wrote: > On Tue, Oct 29, 2013 at 01:13:20PM +0100, Daniel Kiper wrote: > > On Thu, Oct 24, 2013 at 12:47:39PM +0000, OASIS Issues Tracker wrote: > > > Timer/Clock device > > > ------------------ > > > > > > Key: VIRTIO-45 > > > URL: http://tools.oasis-open.org/issues/browse/VIRTIO-45 > > > Project: OASIS Virtual I/O Device (VIRTIO) TC > > > Issue Type: New Feature > > > Reporter: Daniel Kiper > > > > > > > > > There is no timer/clock device in VIRTIO specification. Maybe we should consider addition of one. > > > > General idea is that if we build hardware independent block, network > > and other devices then maybe it makes sens to build hardware independent > > timer/clock. As I know exiting hardware timers/clocks are quite difficult > > to emulate properly in virtual environments. Most problems are solved > > right now but maybe we should still consider addition of timer/clock spec > > in new VIRTIO spec. This way we will make stable time source for virtual > > guests without issues and limitations imposed by existing and new relevant > > hardware devices. > > Do we want that kind of device in VIRTIO spec? VRITIO WD01 is about to release. I think that it is good time to discuss this issue right now. I am looking for your comments on it. Daniel

Daniel Kiper <daniel.kiper@oracle.com> writes: >> Do we want that kind of device in VIRTIO spec? > > VRITIO WD01 is about to release. I think that it is good > time to discuss this issue right now. I am looking for > your comments on it. I have no deep knowledge of how to implement a timer device, so I'd need to study a detailed proposal and take advice from others. Do you have a specific proposal we can throw around? Thanks, Rusty.

On Fri, Jan 17, 2014 at 11:03:30AM +1030, Rusty Russell wrote: > Daniel Kiper <daniel.kiper@oracle.com> writes: > >> Do we want that kind of device in VIRTIO spec? > > > > VRITIO WD01 is about to release. I think that it is good > > time to discuss this issue right now. I am looking for > > your comments on it. > > I have no deep knowledge of how to implement a timer device, so I'd need > to study a detailed proposal and take advice from others. > > Do you have a specific proposal we can throw around? Not yet but I have some loose thoughts. Currently I am checking that current VRITIO spec could be used on Xen without any issue. I think that it is more important right now. Later I am going to work on ballooning stuff. Currently there is no urgent need for timer device. So I think that we could add it in next full release if we do not have a time/resources to work out sensible solution. Daniel

On Fri, Jan 17, 2014 at 01:34:20PM +0100, Daniel Kiper wrote: > On Fri, Jan 17, 2014 at 11:03:30AM +1030, Rusty Russell wrote: > > Daniel Kiper <daniel.kiper@oracle.com> writes: > > >> Do we want that kind of device in VIRTIO spec? > > > > > > VRITIO WD01 is about to release. I think that it is good > > > time to discuss this issue right now. I am looking for > > > your comments on it. > > > > I have no deep knowledge of how to implement a timer device, so I'd need > > to study a detailed proposal and take advice from others. > > > > Do you have a specific proposal we can throw around? > > Not yet but I have some loose thoughts. Currently I am checking that > current VRITIO spec could be used on Xen without any issue. I think > that it is more important right now. Later I am going to work on ballooning > stuff. Currently there is no urgent need for timer device. So I think > that we could add it in next full release if we do not have a time/resources > to work out sensible solution. Rusty, could we reserve an ID for it as we discussed once? If you wish I could post relevant patch. Daniel

Daniel Kiper <daniel.kiper@oracle.com> writes: > On Fri, Jan 17, 2014 at 01:34:20PM +0100, Daniel Kiper wrote: >> On Fri, Jan 17, 2014 at 11:03:30AM +1030, Rusty Russell wrote: >> > Daniel Kiper <daniel.kiper@oracle.com> writes: >> > >> Do we want that kind of device in VIRTIO spec? >> > > >> > > VRITIO WD01 is about to release. I think that it is good >> > > time to discuss this issue right now. I am looking for >> > > your comments on it. >> > >> > I have no deep knowledge of how to implement a timer device, so I'd need >> > to study a detailed proposal and take advice from others. >> > >> > Do you have a specific proposal we can throw around? >> >> Not yet but I have some loose thoughts. Currently I am checking that >> current VRITIO spec could be used on Xen without any issue. I think >> that it is more important right now. Later I am going to work on ballooning >> stuff. Currently there is no urgent need for timer device. So I think >> that we could add it in next full release if we do not have a time/resources >> to work out sensible solution. > > Rusty, could we reserve an ID for it as we discussed once? > If you wish I could post relevant patch. > > Daniel Yes, please do. Then OPEN VIRTIO-45 and attach the URL of the archived mail to it. Thanks! Rusty.