Re: [Xen-devel] [PATCH V3 28/29] x86/vvtd: Add queued invalidation (QI) support

[Chao Gao <chao.gao@xxxxxxxxx>]

On Mon, Oct 23, 2017 at 09:57:16AM +0100, Roger Pau Monné wrote:
>On Mon, Oct 23, 2017 at 03:50:24PM +0800, Chao Gao wrote:
>> On Fri, Oct 20, 2017 at 12:20:06PM +0100, Roger Pau Monné wrote:
>> >On Thu, Sep 21, 2017 at 11:02:09PM -0400, Lan Tianyu wrote:
>> >> From: Chao Gao <chao.gao@xxxxxxxxx>
>> >> +    }
>> >> +
>> >> +    unmap_guest_page((void*)qinval_page);
>> >> +    return ret;
>> >> +
>> >> + error:
>> >> +    unmap_guest_page((void*)qinval_page);
>> >> +    gdprintk(XENLOG_ERR, "Internal error in Queue Invalidation.\n");
>> >> +    domain_crash(vvtd->domain);
>> >
>> >Do you really need to crash the domain in such case?
>> 
>> We reach here when guest requests some operations vvtd doesn't claim
>> supported or emulated. I am afraid it also can be triggered by guest.
>> How about ignoring the invalidation request?
>
>What would real hardware do in such case?

After reading the spec again, I think hardware may generate a fault
event, seeing VT-d spec 10.4.9 Fault Status Register: 
Hardware detected an error associated with the invalidation queue. This
could be due to either a hardware error while fetching a descriptor from
the invalidation queue, or hardware detecting an erroneous or invalid
descriptor in the invalidation queue. At this time, a fault event may be
generated based on the programming of the Fault Event Control register

Thanks
Chao

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel