> From: Xu, Quan > Sent: Tuesday, April 19, 2016 9:28 PM > > On April 19, 2016 2:46pm, Tian, Kevin <kevin.t...@intel.com> wrote: > > > From: Quan Xu > > > Sent: Monday, April 18, 2016 10:00 PM > > > > > > While grant table is unmapping, the domain (with the exception of the > > > > unmapping -> unmapped. > > > > A slightly different take, IMO the hypercall is not returned, so it is DOING.
More accurately you might want to say "When a granted page is being unmapped". > > > > > hardware domain) may be crashed due to IOMMU mapping and unmapping > > > failures, and then it is unnecessary to flush specified CPUs' TLBs. > > > > Above description is not complete. You said "with the exception of the > > hardware domain". Then people will ask whether 'unnecessary' > > is also true for hardware domain (if not restriction for hardware domain to > > invoke those interfaces). > > > > Make sense. > Could I only drop '(with the exception of the hardware domain)'? > Still didn't get the rationale here. After another look at the patch, actually I'm even not sure how the change is related to the purpose of this patch series. Also the change is opposite to your comment - you limit flush to crashed domain instead: > > -fault: > > - gnttab_flush_tlb(current->domain); > > + fault: > > + if ( current->domain->is_shut_down ) > > + gnttab_flush_tlb(current->domain); Thanks Kevin _______________________________________________ Xen-devel mailing list Xen-devel@lists.xen.org http://lists.xen.org/xen-devel
