Techwizz/kernel-fxtec-pro1x
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

kvm: search_memslots: add simple LRU memslot caching

Browse source 
In typical guest boot workload only 2-3 memslots are used
extensively, and at that it's mostly the same memslot
lookup operation.

Adding LRU cache improves average lookup time from
46 to 28 cycles (~40%) for this workload.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
This commit is contained in:
Igor Mammedov 2014-12-01 17:29:25 +00:00 committed by Paolo Bonzini
parent 7f379cff11
commit d4ae84a02b
1 changed files with 10 additions and 2 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

12
include/linux/kvm_host.h
View file

@ -353,6 +353,7 @@ struct kvm_memslots {
struct kvm_memory_slot memslots[KVM_MEM_SLOTS_NUM];
/* The mapping table from slot id to the index in memslots[]. */
short id_to_index[KVM_MEM_SLOTS_NUM];
atomic_t lru_slot;
};
struct kvm {
@ -790,13 +791,20 @@ static inline void kvm_guest_exit(void)
static inline struct kvm_memory_slot *
search_memslots(struct kvm_memslots *slots, gfn_t gfn)
{
struct kvm_memory_slot *memslot;
int slot = atomic_read(&slots->lru_slot);
struct kvm_memory_slot *memslot = &slots->memslots[slot];
kvm_for_each_memslot(memslot, slots)
if (gfn >= memslot->base_gfn &&
gfn < memslot->base_gfn + memslot->npages)
return memslot;
kvm_for_each_memslot(memslot, slots)
if (gfn >= memslot->base_gfn &&
gfn < memslot->base_gfn + memslot->npages) {
atomic_set(&slots->lru_slot, memslot - slots->memslots);
return memslot;
}
return NULL;
}