On Wed, Sep 05, 2012 at 06:12:04PM +0300, Avi Kivity wrote: > On 09/05/2012 06:09 PM, Michael S. Tsirkin wrote: > > On Wed, Sep 05, 2012 at 05:51:53PM +0300, Michael S. Tsirkin wrote: > >> On Wed, Sep 05, 2012 at 05:35:43PM +0300, Avi Kivity wrote: > >> > On 08/22/2012 03:41 AM, Michael S. Tsirkin wrote: > >> > >> > >> > >> I assumed you were pointing out the level vs edge interaction. If we > >> > >> call that a userspace bug, I can just drop this. Thanks, > >> > >> > >> > >> Alex > >> > > > >> > > level is userspace bug I think :) > >> > > >> > I don't see how it's a bug. Suppose we have a vfio device that shares a > >> > gsi with an emulated device. The emulated device naturally uses > >> > KVM_IRQ_LINE (it has no need to re-sample on ADN), while vfio naturally > >> > has to use irqfd. > >> > >> Absolutely. But vfio needs to use irqfd with the new flag. > >> Using existing irqfd for level is a bug. > >> > >> > Note one would expect that each irqfd gets its own irq source id, since > >> > they are all independent level sources. The reason they don't is that > >> > we shut them down anyway and let the sources re-trigger (it is more > >> > accurate to say that they have no irq source id, but that would just > >> > muddle the implementation). > >> > > >> > Alex, if the conclusion is that we do need this patch, then please add a > >> > comment explaining why we can share the source id among all irqfd users. > > > > Something along the lines of > > > > /* > > For resample irqfds, level is a logical OR of all inputs; > > to support this, track state for RESAMPLE irqfds separately > > from userspace. We do not need to track state for each input since > > they are all deasserted at the same time, before resampling. > > */ > > Well the comment style is wrong. Ouch. > To expand a little more, irqfd only sends assert events, so assigning > the level is equivalent to an OR. Clearing an resampling simply builds > the state again. > > btw, there can be other irq source IDs if the lines are shared with the > PIT or kvm assigned devices. Nod. > -- > error compiling committee.c: too many arguments to function -- 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
