KVM: mmu_shrink: kvm_mmu_zap_page requires slots_lock to be held

Original-Commit-Hash: 64f6a0c041bd8fc100a0d655058bdbc31feda03c

kvm_mmu_zap_page() needs slots lock held (rmap_remove->gfn_to_memslot,
for example).

Since kvm_lock spinlock is held in mmu_shrink(), do a non-blocking
down_read_trylock().

Untested.

Signed-off-by: Avi Kivity <avi@qumranet.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c
index 7e7c3969f7a2d01f0f9d2d2542b046fc9d0aa157..c640b2f96c860874c03694ec922a036a164d7ada 100644 (file)
--- a/arch/x86/kvm/mmu.c
+++ b/arch/x86/kvm/mmu.c
@@ -1968,6 +1968,8 @@ static int mmu_shrink(int nr_to_scan, gfp_t gfp_mask)
        list_for_each_entry(kvm, &vm_list, vm_list) {
                int npages;
 
+               if (!down_read_trylock(&kvm->slots_lock))
+                       continue;
                spin_lock(&kvm->mmu_lock);
                npages = kvm->arch.n_alloc_mmu_pages -
                         kvm->arch.n_free_mmu_pages;
@@ -1980,6 +1982,7 @@ static int mmu_shrink(int nr_to_scan, gfp_t gfp_mask)
                nr_to_scan--;
 
                spin_unlock(&kvm->mmu_lock);
+               up_read(&kvm->slots_lock);
        }
        if (kvm_freed)
                list_move_tail(&kvm_freed->vm_list, &vm_list);