[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [PATCH 1/3] VMX: Properly adjuest the status of pi descriptor

> -----Original Message-----
> From: Tian, Kevin
> Sent: Monday, May 23, 2016 1:15 PM
> To: Wu, Feng <feng.wu@xxxxxxxxx>; xen-devel@xxxxxxxxxxxxx
> Cc: keir@xxxxxxx; jbeulich@xxxxxxxx; andrew.cooper3@xxxxxxxxxx;
> george.dunlap@xxxxxxxxxxxxx; dario.faggioli@xxxxxxxxxx;
> konrad.wilk@xxxxxxxxxx
> Subject: RE: [PATCH 1/3] VMX: Properly adjuest the status of pi descriptor
> > From: Wu, Feng
> > Sent: Friday, May 20, 2016 4:54 PM
> >
> > When the last assigned device is dettached from the domain, all
> > the PI related hooks are removed then, however, the vCPU can be
> > blocked, switched to another pCPU, etc, all without the aware of
> > PI. After the next time we attach another device to the domain,
> > which makes the PI realted hooks avaliable again, the status
> > of the pi descriptor is not true, we need to properly adjust
> > it.
> Instead of adjusting pi descriptor in multiple places, can we
> simply reset the status (including removing from block list)
> right when hooks are removed at deattach?

I also thought about this idea before, the problem is that when
the hooks are being removed at the pci device's deattachment,
the hooks may be running at the same time, which may cause
problems, such as, after we have removed the vCPU from the
blocking list, vmx_vcpu_block() (which has been running before the
hooks were removed and not finished yet.) adds a vCPU to the same
blocking list, then this vCPU will remain in the list after the device is
deattached from the domain. At the same reason, if we change PI desc
in vmx_pi_hooks_deassign() while it is being used in the hooks, it
may cause problems.


Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.