In theory, we should inject data_page_not_present, While seems data_page_not_present is more expense than dtlb_miss Data_page_not_present executes rsm psr.dt ( which is not needed for native linux) and ptc.l ( which is not needed in virtual environment).
Current implementation can achieve a little better performance. Seems same thing happens to windows. - Anthony >-----Original Message----- >From: [EMAIL PROTECTED] [mailto:xen-ia64-devel- >[EMAIL PROTECTED] On Behalf Of Tristan Gingold >Sent: 2007年9月30日 23:29 >To: Xen-ia64-devel >Subject: [Xen-ia64-devel] Question in vmx_fault.c > >Hi, > >in vmx_hpw_miss (DTLB part): > if (!guest_vhpt_lookup(vhpt_adr, &pteval)) { > /* VHPT successfully read. */ > if (!(pteval & _PAGE_P)) { > if (vpsr.ic) { > vcpu_set_isr(v, misr.val); > dtlb_fault(v, vadr); /* XXXXXXXXXXXX */ > return IA64_FAULT; > } else { > nested_dtlb(v); > return IA64_FAULT; > } > >Shouldn't we call data_page_not_present instead ? > >Tristan. > >_______________________________________________ >Xen-ia64-devel mailing list >[email protected] >http://lists.xensource.com/xen-ia64-devel _______________________________________________ Xen-ia64-devel mailing list [email protected] http://lists.xensource.com/xen-ia64-devel
