Hi, Tim, Just a curious question. Any reason why sh_page_fault can''t benefit vtlb_lookup to skip heavy-weight guest_walk_tables, like other places like sh_gva_to_gfn? Thanks, Kevin _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
At 18:22 +0800 on 05 Dec (1196878922), Tian, Kevin wrote:> Hi, Tim, > Just a curious question. Any reason why sh_page_fault can''t > benefit vtlb_lookup to skip heavy-weight guest_walk_tables, like other > places like sh_gva_to_gfn?For cases when we go on to make the shadow pagetables, we need the full walk available so we know which MFNs to shadow. I don''t know whether it would be worth adding a vtlb lookup for the real-fault case; the fast-path code for not-present and MMIO catches a lot of genuine faults already. We should definitely add the walk result to the vTLB in any case. Thanks, Tim. -- Tim Deegan <Tim.Deegan@citrix.com> Principal Software Engineer, Citrix Systems. [Company #5334508: XenSource UK Ltd, reg''d c/o EC2Y 5EB, UK.] _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
>From: Tim Deegan [mailto:Tim.Deegan@citrix.com] >Sent: 2007年12月5日 18:35 > >At 18:22 +0800 on 05 Dec (1196878922), Tian, Kevin wrote: >> Hi, Tim, >> Just a curious question. Any reason why sh_page_fault can''t >> benefit vtlb_lookup to skip heavy-weight guest_walk_tables, >like other >> places like sh_gva_to_gfn? > >For cases when we go on to make the shadow pagetables, we need the full >walk available so we know which MFNs to shadow. > >I don''t know whether it would be worth adding a vtlb lookup for the >real-fault case; the fast-path code for not-present and MMIO catches a >lot of genuine faults already. We should definitely add the >walk result >to the vTLB in any case. > >Thanks, > >Tim. >Understand. fast-path should be enough without vtlb help. Thanks, Kevin _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
>From: Tian, Kevin >Sent: 2007年12月6日 9:41 > >>From: Tim Deegan [mailto:Tim.Deegan@citrix.com] >>Sent: 2007年12月5日 18:35 >> >>At 18:22 +0800 on 05 Dec (1196878922), Tian, Kevin wrote: >>> Hi, Tim, >>> Just a curious question. Any reason why sh_page_fault can''t >>> benefit vtlb_lookup to skip heavy-weight guest_walk_tables, >>like other >>> places like sh_gva_to_gfn? >> >>For cases when we go on to make the shadow pagetables, we >need the full >>walk available so we know which MFNs to shadow. >> >>I don''t know whether it would be worth adding a vtlb lookup for the >>real-fault case; the fast-path code for not-present and MMIO catches a >>lot of genuine faults already. We should definitely add the >>walk result >>to the vTLB in any case. >> >>Thanks, >> >>Tim. >> > >Understand. fast-path should be enough without vtlb help. > >Thanks, >Kevin >BTW, is it cleaner to move vtlb_insert into guest_walk_tables, which is more natural and save the effort to update all invocation points? Thanks, Kevin _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel