On Sat, Aug 07, 2021, Paolo Bonzini wrote:
> @@ -377,9 +377,9 @@ TRACE_EVENT(
> ),
>
> TP_fast_assign(
> - __entry->gfn = addr >> PAGE_SHIFT;
> - __entry->pfn = pfn | (__entry->gfn & (KVM_PAGES_PER_HPAGE(level) - 1));
> - __entry->level = level;
> + __entry->gfn = fault->addr >> PAGE_SHIFT;
Eww. The existing code also bastardizes addr vs. gpa, but this just looks even
more wrong because we have fault->gfn.
Maybe do this as a prep patch at the beginning of the series? And then use
fault->gfn directly.
diff --git a/arch/x86/kvm/mmu/paging_tmpl.h b/arch/x86/kvm/mmu/paging_tmpl.h
index 7d03e9b7ccfa..b159749300b5 100644
--- a/arch/x86/kvm/mmu/paging_tmpl.h
+++ b/arch/x86/kvm/mmu/paging_tmpl.h
@@ -725,7 +725,7 @@ static int FNAME(fetch)(struct kvm_vcpu *vcpu, gpa_t addr,
level = kvm_mmu_hugepage_adjust(vcpu, gw->gfn, max_level, &pfn,
huge_page_disallowed, &req_level);
- trace_kvm_mmu_spte_requested(addr, gw->level, pfn);
+ trace_kvm_mmu_spte_requested(gw->gfn << PAGE_SHIFT, gw->level, pfn);
for (; shadow_walk_okay(&it); shadow_walk_next(&it)) {
clear_sp_write_flooding_count(it.sptep);
> + __entry->pfn = fault->pfn | (__entry->gfn & (KVM_PAGES_PER_HPAGE(fault->goal_level) - 1));
Similar thing here, it could use fault->gfn directly.
> + __entry->level = fault->goal_level;
> ),
>
> TP_printk("gfn %llx pfn %llx level %d",
