linux-kernel - Re: [PATCH 5/6] KVM: x86/mmu: Protect kvm->memslots with a mutex

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Wed, 28 Apr 2021 19:46:39 +0200
From:   Paolo Bonzini <pbonzini@...hat.com>
To:     Ben Gardon <bgardon@...gle.com>
Cc:     LKML <linux-kernel@...r.kernel.org>, kvm <kvm@...r.kernel.org>,
        Peter Xu <peterx@...hat.com>,
        Sean Christopherson <seanjc@...gle.com>,
        Peter Shier <pshier@...gle.com>,
        Junaid Shahid <junaids@...gle.com>,
        Jim Mattson <jmattson@...gle.com>,
        Yulei Zhang <yulei.kernel@...il.com>,
        Wanpeng Li <kernellwp@...il.com>,
        Vitaly Kuznetsov <vkuznets@...hat.com>,
        Xiao Guangrong <xiaoguangrong.eric@...il.com>
Subject: Re: [PATCH 5/6] KVM: x86/mmu: Protect kvm->memslots with a mutex

On 28/04/21 18:40, Ben Gardon wrote:
> On Tue, Apr 27, 2021 at 11:25 PM Paolo Bonzini <pbonzini@...hat.com> wrote:
>>
>> On 28/04/21 00:36, Ben Gardon wrote:
>>> +void kvm_arch_assign_memslots(struct kvm *kvm, int as_id,
>>> +                          struct kvm_memslots *slots)
>>> +{
>>> +     mutex_lock(&kvm->arch.memslot_assignment_lock);
>>> +     rcu_assign_pointer(kvm->memslots[as_id], slots);
>>> +     mutex_unlock(&kvm->arch.memslot_assignment_lock);
>>> +}
>>
>> Does the assignment also needs the lock, or only the rmap allocation?  I
>> would prefer the hook to be something like kvm_arch_setup_new_memslots.
> 
> The assignment does need to be under the lock to prevent the following race:
> 1. Thread 1 (installing a new memslot): Acquires memslot assignment
> lock (or perhaps in this case rmap_allocation_lock would be more apt.)
> 2. Thread 1: Check alloc_memslot_rmaps (it is false)
> 3. Thread 1: doesn't allocate memslot rmaps for new slot
> 4. Thread 1: Releases memslot assignment lock
> 5. Thread 2 (allocating a shadow root): Acquires memslot assignment lock
> 6. Thread 2: Sets alloc_memslot_rmaps = true
> 7. Thread 2: Allocates rmaps for all existing slots
> 8. Thread 2: Releases memslot assignment lock
> 9. Thread 2: Sets shadow_mmu_active = true
> 10. Thread 1: Installs the new memslots
> 11. Thread 3: Null pointer dereference when trying to access rmaps on
> the new slot.

... because thread 3 would be under mmu_lock and therefore cannot 
allocate the rmap itself (you have to do it in mmu_alloc_shadow_roots, 
as in patch 6).

Related to this, your solution does not have to protect kvm_dup_memslots 
with the new lock, because the first update of the memslots will not go 
through kvm_arch_prepare_memory_region but it _will_ go through 
install_new_memslots and therefore through the new hook.  But overall I 
think I'd prefer to have a kvm->slots_arch_lock mutex in generic code, 
and place the call to kvm_dup_memslots and 
kvm_arch_prepare_memory_region inside that mutex.

That makes the new lock decently intuitive, and easily documented as 
"Architecture code can use slots_arch_lock if the contents of struct 
kvm_arch_memory_slot needs to be written outside 
kvm_arch_prepare_memory_region.  Unlike slots_lock, slots_arch_lock can 
be taken inside a ``kvm->srcu`` read-side critical section".

I admit I haven't thought about it very thoroughly, but if something 
like this is enough, it is relatively pretty:

diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index 9b8e30dd5b9b..6e5106365597 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -1333,6 +1333,7 @@ static struct kvm_memslots 
*install_new_memslots(struct kvm *kvm,

  	rcu_assign_pointer(kvm->memslots[as_id], slots);

+	mutex_unlock(&kvm->slots_arch_lock);
  	synchronize_srcu_expedited(&kvm->srcu);

  	/*
@@ -1399,6 +1398,7 @@ static int kvm_set_memslot(struct kvm *kvm,
  	struct kvm_memslots *slots;
  	int r;

+	mutex_lock(&kvm->slots_arch_lock);
  	slots = kvm_dup_memslots(__kvm_memslots(kvm, as_id), change);
  	if (!slots)
  		return -ENOMEM;
@@ -1427,6 +1427,7 @@ static int kvm_set_memslot(struct kvm *kvm,
  		 *	- kvm_is_visible_gfn (mmu_check_root)
  		 */
  		kvm_arch_flush_shadow_memslot(kvm, slot);
+		mutex_lock(&kvm->slots_arch_lock);
  	}

  	r = kvm_arch_prepare_memory_region(kvm, new, mem, change);

It does make the critical section a bit larger, so that the 
initialization of the shadow page (which is in KVM_RUN context) contends 
with slightly more code than necessary.  However it's all but a 
performance critical situation, as it will only happen just once per VM.

WDYT?

Paolo

> Putting the assignment under the lock prevents 5-8 from happening
> between 2 and 10.
> 
> I'm open to other ideas as far as how to prevent this race though. I
> admit this solution is not the most elegant looking.