[PATCH v4 0/2] KVM: arm64: vgic: Fix racy LPI release and re-registration handling

[PATCH v4 0/2] KVM: arm64: vgic: Fix racy LPI release and re-registration handling

[Carlos López]

Fix a couple of potential issues that could arise from racy LPI release
and re-registration for the same INTID.

The issue fixed in patch 1 can manifest itself through either a
leaked LPI structure, or a prematurely deleted LPI. The issue fixed
in patch 2 could materialize as a spurious -ENOMEM failure when
registering an LPI.

v4:
* Add __GFP_ACCOUNT to patch 2 (Sashiko).

v3:
* Use refcount_dec_and_lock_irqsave() instead of unconditionally
  grabbing the xarray lock in patch 1.
* Add patch 2.

v2:
* Address Sashiko's review. Fix the direct release path by decrementing
  the refcount under the xarray spinlock, preventing a UAF that would
  have been introduced in v1.


Carlos López (2):
  KVM: arm64: vgic: Fix race between LPI release and re-registration
  KVM: arm64: vgic: Mitigate potential LPI registration failure

 arch/arm64/kvm/vgic/vgic-its.c | 11 ++++++++++-
 arch/arm64/kvm/vgic/vgic.c     | 10 ++++++----
 2 files changed, 16 insertions(+), 5 deletions(-)


base-commit: 1ee27dacbe5dc4def481794d899d67b0d4570094
-- 
2.51.0

[PATCH v4 1/2] KVM: arm64: vgic: Fix race between LPI release and re-registration

[Carlos López]

Fix a potential race between decrementing an LPI's reference count and
evicting that structure from the LPI xarray.

LPI structures are maintained in the VGIC LPI xarray (dist->lpi_xa).
When the reference count of an LPI structure drops to zero,
vgic_release_lpi_locked() removes the structure from the xarray and
frees it under the xarray lock.

However, the release of an LPI can race with a concurrent LPI
re-registration with the same INTID via vgic_add_lpi() on another CPU,
since the reference count drop and the xarray eviction are not performed
in a single atomic step. This can happen e.g. if the guest issues a
DISCARD while the LPI is still referenced from a vCPU's active-pending
list (ap_list), and the same INTID is re-mapped via MAPTI.

Particularly, vgic_release_lpi_locked() is called from two distinct
paths: direct release via vgic_put_irq(), and deferred release via
vgic_release_deleted_lpis(). During direct release, the issue can result
in deleting a newly registered LPI from the xarray:

  CPU0 (Releasing LPI)                    CPU1 (Adding new LPI)
  ====================                    =====================
  vgic_put_irq()
      __vgic_put_irq()
          refcount_dec_and_test()
                                          vgic_add_lpi()
                                              xa_lock_irqsave()
                                              old_irq = xa_load(.., intid)
                                              vgic_try_get_irq_ref(old_irq) == false
                        new IRQ inserted -->  __xa_store(.., intid, ..)
                                              xa_unlock_irqrestore()
  xa_lock_irqsave();
  vgic_release_lpi_locked()
      __xa_erase(.., irq->intid)   <-- BUG: new IRQ is erased
      kfree_rcu(old_irq)

During the deferred release path, the old IRQ can be leaked:

  CPU0 (Releasing LPI)                    CPU1 (Adding new LPI)
  ====================                    =====================
  vgic_put_irq_norelease()
      __vgic_put_irq()
          refcount_dec_and_test()
      irq->pending_release = true
                                          vgic_add_lpi()
                                              xa_lock_irqsave()
                                              old_irq = xa_load(.., intid)
                                              vgic_try_get_irq_ref(oldirq) == false
                 BUG: old IRQ overwritten --> __xa_store(.., intid, ..)
                                              xa_unlock_irqrestore()

  vgic_release_deleted_lpis()
      xa_lock_irqsave()
      xa_for_each() { .. } <-- old IRQ with pending_release = true
                               is gone, so it cannot be released

To fix the direct release path, move the reference count drop inside
the xarray lock, making sure that vgic_add_lpi() never encounters the
to-be-released LPI.

To fix the deferred release path, since the refcount drop must happen
under a raw spinlock, the same solution does not work. Instead, update
vgic_add_lpi(), so that if it evicts a non-NULL refcount=0 LPI from the
xarray, it takes on the responsibility of releasing it. If this happens,
vgic_release_deleted_lpis() will iterate the xarray normally and will
simply not find the already released structure.

Reported-by: Claude:claude-opus-4-6
Fixes: 3a08a6ca7c37 ("KVM: arm64: vgic-v3: Use bare refcount for VGIC LPIs")
Fixes: d54594accf73 ("KVM: arm64: vgic-v3: Erase LPIs from xarray outside of raw spinlocks")
Signed-off-by: Carlos López <clopez@suse.de>
---
 arch/arm64/kvm/vgic/vgic-its.c | 10 +++++++++-
 arch/arm64/kvm/vgic/vgic.c     | 10 ++++++----
 2 files changed, 15 insertions(+), 5 deletions(-)

diff --git a/arch/arm64/kvm/vgic/vgic-its.c b/arch/arm64/kvm/vgic/vgic-its.c
index 67d107e9a77d..577286069368 100644
--- a/arch/arm64/kvm/vgic/vgic-its.c
+++ b/arch/arm64/kvm/vgic/vgic-its.c
@@ -116,7 +116,15 @@ static struct vgic_irq *vgic_add_lpi(struct kvm *kvm, u32 intid,
 		kfree(irq);
 		irq = oldirq;
 	} else {
-		ret = xa_err(__xa_store(&dist->lpi_xa, intid, irq, 0));
+		/*
+		 * The entry is either empty or contains a dead LPI (refcount=0)
+		 * from the deferred release path, pending cleanup by
+		 * vgic_release_deleted_lpis(). Evict and free it if present.
+		 */
+		oldirq = __xa_store(&dist->lpi_xa, intid, irq, 0);
+		ret = xa_err(oldirq);
+		if (!ret && oldirq)
+			kfree_rcu(oldirq, rcu);
 	}
 
 	xa_unlock_irqrestore(&dist->lpi_xa, flags);
diff --git a/arch/arm64/kvm/vgic/vgic.c b/arch/arm64/kvm/vgic/vgic.c
index 5a4768d8cd4f..cc09e0c45b46 100644
--- a/arch/arm64/kvm/vgic/vgic.c
+++ b/arch/arm64/kvm/vgic/vgic.c
@@ -167,12 +167,14 @@ void vgic_put_irq(struct kvm *kvm, struct vgic_irq *irq)
 		guard(spinlock_irqsave)(&dist->lpi_xa.xa_lock);
 	}
 
-	if (!__vgic_put_irq(kvm, irq))
+	if (!irq_is_lpi(kvm, irq->intid))
 		return;
 
-	xa_lock_irqsave(&dist->lpi_xa, flags);
-	vgic_release_lpi_locked(dist, irq);
-	xa_unlock_irqrestore(&dist->lpi_xa, flags);
+	if (refcount_dec_and_lock_irqsave(&irq->refcount,
+					  &dist->lpi_xa.xa_lock, &flags)) {
+		vgic_release_lpi_locked(dist, irq);
+		xa_unlock_irqrestore(&dist->lpi_xa, flags);
+	}
 }
 
 static void vgic_release_deleted_lpis(struct kvm *kvm)
-- 
2.51.0

[PATCH v4 2/2] KVM: arm64: vgic: Mitigate potential LPI registration failure

[Carlos López]

Mitigate a potential failure when inserting a new LPI into the VGIC LPI
xarray.

When vgic_add_lpi() is preparing to register a new LPI, it pre-allocates
an xarray entry using xa_reserve_irq(), so that it can later perform the
insertion under the xarray lock without allocating.

However, since xa_reserve_irq() is called before acquiring such lock,
there is a potential race where xa_reserve_irq() observes a populated
entry, thus not performing the allocation, and another CPU removes that
entry before the xarray lock is grabbed to perform the insertion.

  CPU0 (Adding new LPI)                      CPU1 (Releasing LPI)
  =====================                      ===================
  vgic_add_lpi()
      /* Entry populated, does not allocate */
      xa_reserve_irq(.., intid, ..)
                                            vgic_release_deleted_lpis()
                                                xa_lock_irqsave()
                                                vgic_release_lpi_locked()
                          xarray node freed --> __xa_erase(.., intid)
                                                xa_unlock_irqrestore()
      xa_lock_irqsave()
      xa_load(.., intid) == NULL
      vgic_try_get_irq_ref(NULL) == false
      __xa_store(.., intid, irq, 0) <-- xarray node was freed, gfp=0
                                        cannot allocate, returns -ENOMEM

This can happen e.g. if the guest issues a DISCARD while the LPI is
still referenced from a vCPU's active-pending list (ap_list), and the
same INTID is re-mapped via MAPTI.

Mitigate this by passing GFP_NOWAIT to __xa_store(), so that the
allocation can happen under the lock in the rare case that this
condition is hit. Add __GFP_ACCOUNT as well to match xa_reserve_irq()'s
flags.

Reported-by: Sashiko <sashiko-bot@kernel.org>
Fixes: 1d6f83f60f79 ("KVM: arm64: vgic: Store LPIs in an xarray")
Signed-off-by: Carlos López <clopez@suse.de>
---
 arch/arm64/kvm/vgic/vgic-its.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/arch/arm64/kvm/vgic/vgic-its.c b/arch/arm64/kvm/vgic/vgic-its.c
index 577286069368..327418477c51 100644
--- a/arch/arm64/kvm/vgic/vgic-its.c
+++ b/arch/arm64/kvm/vgic/vgic-its.c
@@ -121,7 +121,8 @@ static struct vgic_irq *vgic_add_lpi(struct kvm *kvm, u32 intid,
 		 * from the deferred release path, pending cleanup by
 		 * vgic_release_deleted_lpis(). Evict and free it if present.
 		 */
-		oldirq = __xa_store(&dist->lpi_xa, intid, irq, 0);
+		oldirq = __xa_store(&dist->lpi_xa, intid, irq,
+				    GFP_NOWAIT | __GFP_ACCOUNT);
 		ret = xa_err(oldirq);
 		if (!ret && oldirq)
 			kfree_rcu(oldirq, rcu);
-- 
2.51.0

Re: [PATCH v4 1/2] KVM: arm64: vgic: Fix race between LPI release and re-registration

[Oliver Upton]

Hi Carlos,

Please wait a few days between spins, we were in the middle of a
conversation and I didn't have a chance to reply to you since yesterday.

On Fri, Jul 03, 2026 at 02:26:00PM +0200, Carlos López wrote:
> diff --git a/arch/arm64/kvm/vgic/vgic-its.c b/arch/arm64/kvm/vgic/vgic-its.c
> index 67d107e9a77d..577286069368 100644
> --- a/arch/arm64/kvm/vgic/vgic-its.c
> +++ b/arch/arm64/kvm/vgic/vgic-its.c
> @@ -116,7 +116,15 @@ static struct vgic_irq *vgic_add_lpi(struct kvm *kvm, u32 intid,
>  		kfree(irq);
>  		irq = oldirq;
>  	} else {
> -		ret = xa_err(__xa_store(&dist->lpi_xa, intid, irq, 0));
> +		/*
> +		 * The entry is either empty or contains a dead LPI (refcount=0)
> +		 * from the deferred release path, pending cleanup by
> +		 * vgic_release_deleted_lpis(). Evict and free it if present.
> +		 */
> +		oldirq = __xa_store(&dist->lpi_xa, intid, irq, 0);
> +		ret = xa_err(oldirq);
> +		if (!ret && oldirq)
> +			kfree_rcu(oldirq, rcu);

We need to assert the assumption here that the evicted IRQ has a nonzero
refcount and is pending release. Otherwise I feel like possibly leaking
memory is better than potentially introducing a UAF.

Untested diff below.

Thanks,
Oliver

diff --git a/arch/arm64/kvm/vgic/vgic-its.c b/arch/arm64/kvm/vgic/vgic-its.c
index 4477f870c7b3..621a83399184 100644
--- a/arch/arm64/kvm/vgic/vgic-its.c
+++ b/arch/arm64/kvm/vgic/vgic-its.c
@@ -116,18 +116,28 @@ static struct vgic_irq *vgic_add_lpi(struct kvm *kvm, u32 intid,
 		kfree(irq);
 		irq = oldirq;
 	} else {
-		ret = xa_err(__xa_store(&dist->lpi_xa, intid, irq, 0));
-	}
-
-	xa_unlock_irqrestore(&dist->lpi_xa, flags);
+		oldirq = __xa_store(&dist->lpi_xa, intid, irq, 0);
+		ret = xa_err(oldirq);
+		if (ret) {
+			xa_unlock_irqrestore(&dist->lpi_xa, flags);
+			kfree(irq);
 
-	if (ret) {
-		xa_release(&dist->lpi_xa, intid);
-		kfree(irq);
+			return ERR_PTR(ret);
+		}
 
-		return ERR_PTR(ret);
+		/*
+		 * The expectation is that only LPIs which are pending release
+		 * could possibly be evicted by storing the new entry. Free the
+		 * old IRQ only if this is the case, as otherwise someone else
+		 * could still hold a pointer on the outgoing IRQ.
+		 */
+		if (oldirq && !WARN_ON_ONCE(refcount_read(&oldirq->refcount) &&
+					    oldirq->pending_release))
+			kfree_rcu(oldirq, rcu);
 	}
 
+	xa_unlock_irqrestore(&dist->lpi_xa, flags);
+
 	/*
 	 * We "cache" the configuration table entries in our struct vgic_irq's.
 	 * However we only have those structs for mapped IRQs, so we read in

Re: [PATCH v4 1/2] KVM: arm64: vgic: Fix race between LPI release and re-registration

[Oliver Upton]

On Fri, Jul 03, 2026 at 03:13:34PM -0700, Oliver Upton wrote:
> Hi Carlos,
> 
> Please wait a few days between spins, we were in the middle of a
> conversation and I didn't have a chance to reply to you since yesterday.
> 
> On Fri, Jul 03, 2026 at 02:26:00PM +0200, Carlos López wrote:
> > diff --git a/arch/arm64/kvm/vgic/vgic-its.c b/arch/arm64/kvm/vgic/vgic-its.c
> > index 67d107e9a77d..577286069368 100644
> > --- a/arch/arm64/kvm/vgic/vgic-its.c
> > +++ b/arch/arm64/kvm/vgic/vgic-its.c
> > @@ -116,7 +116,15 @@ static struct vgic_irq *vgic_add_lpi(struct kvm *kvm, u32 intid,
> >  		kfree(irq);
> >  		irq = oldirq;
> >  	} else {
> > -		ret = xa_err(__xa_store(&dist->lpi_xa, intid, irq, 0));
> > +		/*
> > +		 * The entry is either empty or contains a dead LPI (refcount=0)
> > +		 * from the deferred release path, pending cleanup by
> > +		 * vgic_release_deleted_lpis(). Evict and free it if present.
> > +		 */
> > +		oldirq = __xa_store(&dist->lpi_xa, intid, irq, 0);
> > +		ret = xa_err(oldirq);
> > +		if (!ret && oldirq)
> > +			kfree_rcu(oldirq, rcu);
> 
> We need to assert the assumption here that the evicted IRQ has a nonzero

s/nonzero/zero

Obviously not reading what the hell I'm writing :)

> refcount and is pending release. Otherwise I feel like possibly leaking
> memory is better than potentially introducing a UAF.
> 
> Untested diff below.
> 
> Thanks,
> Oliver
> 
> diff --git a/arch/arm64/kvm/vgic/vgic-its.c b/arch/arm64/kvm/vgic/vgic-its.c
> index 4477f870c7b3..621a83399184 100644
> --- a/arch/arm64/kvm/vgic/vgic-its.c
> +++ b/arch/arm64/kvm/vgic/vgic-its.c
> @@ -116,18 +116,28 @@ static struct vgic_irq *vgic_add_lpi(struct kvm *kvm, u32 intid,
>  		kfree(irq);
>  		irq = oldirq;
>  	} else {
> -		ret = xa_err(__xa_store(&dist->lpi_xa, intid, irq, 0));
> -	}
> -
> -	xa_unlock_irqrestore(&dist->lpi_xa, flags);
> +		oldirq = __xa_store(&dist->lpi_xa, intid, irq, 0);
> +		ret = xa_err(oldirq);
> +		if (ret) {
> +			xa_unlock_irqrestore(&dist->lpi_xa, flags);
> +			kfree(irq);
>  
> -	if (ret) {
> -		xa_release(&dist->lpi_xa, intid);
> -		kfree(irq);
> +			return ERR_PTR(ret);
> +		}
>  
> -		return ERR_PTR(ret);
> +		/*
> +		 * The expectation is that only LPIs which are pending release
> +		 * could possibly be evicted by storing the new entry. Free the
> +		 * old IRQ only if this is the case, as otherwise someone else
> +		 * could still hold a pointer on the outgoing IRQ.
> +		 */
> +		if (oldirq && !WARN_ON_ONCE(refcount_read(&oldirq->refcount) &&
> +					    oldirq->pending_release))

		if (oldirq && !WARN_ON_ONCE(refcount_read(&oldirq->refcount) ||
					    !oldirq->pending_release))

Thanks,
Oliver