In mmu_shrink(), we walk the vm_list to try and free some of the mmu objects. If we did this over and over again, we would unfairly bias the shrinking at the beginning of the list. So, if a KVM instance gets successfully shrunk, then we move it to the end of the vm_list. This keeps things fair. With the global mmu object counter, we no longer need to look at each kvm object on the vm_list during each mmu_shrink() operation. So, move the list manipulation to inside of the loop, and get rid of the 'kvm_freed' variable since we no longer need to store it. This also removes the manipulation of the 'nr_to_scan' variable. It use in here was questionable, especially since it was being decremented even in cases where no scanning was taking place: when building the counter. Interestingly enough, removing it here does not affect the reclaim behavior at all. Signed-off-by: Dave Hansen <dave@xxxxxxxxxxxxxxxxxx> --- linux-2.6.git-dave/arch/x86/kvm/mmu.c | 15 +++++---------- 1 file changed, 5 insertions(+), 10 deletions(-) diff -puN arch/x86/kvm/mmu.c~optimize_shrinker-2 arch/x86/kvm/mmu.c --- linux-2.6.git/arch/x86/kvm/mmu.c~optimize_shrinker-2 2010-06-14 10:24:48.000000000 -0700 +++ linux-2.6.git-dave/arch/x86/kvm/mmu.c 2010-06-14 10:24:48.000000000 -0700 @@ -2951,7 +2951,6 @@ static int shrink_kvm_mmu(struct kvm *kv static int mmu_shrink(int nr_to_scan, gfp_t gfp_mask) { struct kvm *kvm; - struct kvm *kvm_freed = NULL; if (nr_to_scan == 0) goto out; @@ -2959,18 +2958,14 @@ static int mmu_shrink(int nr_to_scan, gf spin_lock(&kvm_lock); list_for_each_entry(kvm, &vm_list, vm_list) { - if (nr_to_scan <= 0) - break; + int freed = shrink_kvm_mmu(kvm, nr_to_scan); + if (!freed) + continue; - shrink_kvm_mmu(kvm, nr_to_scan); - if (!kvm_freed) - kvm_freed = kvm; - nr_to_scan--; + list_move_tail(&kvm->vm_list, &vm_list); + break; } - if (kvm_freed) - list_move_tail(&kvm_freed->vm_list, &vm_list); - spin_unlock(&kvm_lock); out: _ -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html