[PATCH 09/35] KVM: PPC: Make invalidation code more reliable
From: Alexander Graf <hidden>
Date: 2010-08-31 02:32:21
Also in:
kvm
Subsystem:
kernel virtual machine for powerpc (kvm/powerpc), linux for powerpc (32-bit and 64-bit), the rest · Maintainers:
Madhavan Srinivasan, Michael Ellerman, Linus Torvalds
There is a race condition in the pte invalidation code path where we can't be sure if a pte was invalidated already. So let's move the spin lock around to get rid of the race. Signed-off-by: Alexander Graf <redacted> --- arch/powerpc/kvm/book3s_mmu_hpte.c | 14 ++++++++------ 1 files changed, 8 insertions(+), 6 deletions(-)
diff --git a/arch/powerpc/kvm/book3s_mmu_hpte.c b/arch/powerpc/kvm/book3s_mmu_hpte.c
index bd6a767..79751d8 100644
--- a/arch/powerpc/kvm/book3s_mmu_hpte.c
+++ b/arch/powerpc/kvm/book3s_mmu_hpte.c@@ -92,10 +92,6 @@ static void free_pte_rcu(struct rcu_head *head) static void invalidate_pte(struct kvm_vcpu *vcpu, struct hpte_cache *pte) { - /* pte already invalidated? */ - if (hlist_unhashed(&pte->list_pte)) - return; - trace_kvm_book3s_mmu_invalidate(pte); /* Different for 32 and 64 bit */
@@ -103,18 +99,24 @@ static void invalidate_pte(struct kvm_vcpu *vcpu, struct hpte_cache *pte) spin_lock(&vcpu->arch.mmu_lock); + /* pte already invalidated in between? */ + if (hlist_unhashed(&pte->list_pte)) { + spin_unlock(&vcpu->arch.mmu_lock); + return; + } + hlist_del_init_rcu(&pte->list_pte); hlist_del_init_rcu(&pte->list_pte_long); hlist_del_init_rcu(&pte->list_vpte); hlist_del_init_rcu(&pte->list_vpte_long); - spin_unlock(&vcpu->arch.mmu_lock); - if (pte->pte.may_write) kvm_release_pfn_dirty(pte->pfn); else kvm_release_pfn_clean(pte->pfn); + spin_unlock(&vcpu->arch.mmu_lock); + vcpu->arch.hpte_cache_count--; call_rcu(&pte->rcu_head, free_pte_rcu); }
--
1.6.0.2