[PATCH v2] vfio: Use file-based reference counting for KVM

[PATCH v2] vfio: Use file-based reference counting for KVM

[Steffen Eiden]

Replace manual module reference counting with file-based reference
counting for KVM integration. Previously, VFIO used symbol_get() to
obtain function pointers for kvm_get_kvm_safe() and kvm_put_kvm(),
then manually tracked module references through these symbols. This
approach required storing the put_kvm function pointer in each device
and carefully managing symbol references. Remove the put_kvm field in
struct vfio_device as is it no longer used.

Pass struct file pointers instead of struct kvm pointers throughout the
VFIO-KVM interface. This leverages the kernel's existing file reference
counting mechanism via get_file()/get_file_active() and fput(),
eliminating the need for manual module reference tracking. The
file->private_data field provides access to the underlying struct kvm
when needed.

group->kvm and df->kvm hold a reference of their own, taken when the
pointer is stored and dropped when it is overwritten or cleared. They
have to. The kvm-vfio device fd holds a VM reference of its own, so the
VM file can be closed and released while the kvm-vfio device is still
alive and still pointing at it.

kvm->file itself carries no reference, so that it does not pin the VM.
It is only ever read with get_file_active(), which is safe because
kvm_vm_release() clears it, i.e. before the struct file is freed.

This simplifies the code and removes all remaining externally exported
symbols for KVM, paving the path for a second concurrent KVM module.

Suggested-by: Jason Gunthorpe <jgg@nvidia.com>
Co-developed-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Reviewed-by: Jason J. Herne <jjherne@linux.ibm.com>
Signed-off-by: Steffen Eiden <seiden@linux.ibm.com>
---
This is a spin-off for the arm-on-s390 series [1] to address all the
relevant persons w/o polluting their mailbox.

The motivation of this patch is to remove all exported symbols from KVM
to other modules. This patch helps us having multiple KVMs running in
parallel. The arm on s390 series creates a second kvm module (kvm-arm64)
for hosting arm guests on s390 with HW acceleration. For this we need
two concurrent KVM modules. One for s390 (native) and one for arm64
guests.

Steffen

[1] https://lore.kernel.org/all/20260831144802.834315-1-seiden@linux.ibm.com/

Changes in v2:
- Rebase on 7.3-rc1+ and resolve conflict with 9f240376d034 ("s390/pci: Store
  PCI error information for passthrough devices"),
- Fix stub prototype for vfio_group_set_kvm for !CONFIG_VFIO_GROUP
- Link to v1:   https://lore.kernel.org/r/20260812-vfio-v1-1-5cfe0b1fa4e7@linux.ibm.com
---
 arch/x86/include/asm/kvm_page_track.h |  8 ++---
 arch/x86/kvm/Makefile                 |  5 +--
 arch/x86/kvm/mmu/page_track.c         | 20 +++++++-----
 drivers/s390/crypto/vfio_ap_ops.c     | 20 ++++++++----
 drivers/vfio/group.c                  | 11 ++++++-
 drivers/vfio/pci/vfio_pci_zdev.c      |  8 ++++-
 drivers/vfio/vfio.h                   | 12 ++++----
 drivers/vfio/vfio_main.c              | 57 +++++++++++------------------------
 include/linux/kvm_host.h              |  8 +++++
 include/linux/vfio.h                  |  5 ++-
 virt/kvm/kvm_main.c                   | 20 ++++++++++--
 virt/kvm/vfio.c                       | 13 +++++---
 12 files changed, 108 insertions(+), 79 deletions(-)

diff --git a/arch/x86/include/asm/kvm_page_track.h b/arch/x86/include/asm/kvm_page_track.h
index 3d040741044b..046a25c8fe4f 100644
--- a/arch/x86/include/asm/kvm_page_track.h
+++ b/arch/x86/include/asm/kvm_page_track.h
@@ -44,13 +44,13 @@ struct kvm_page_track_notifier_node {
 				    struct kvm_page_track_notifier_node *node);
 };
 
-int kvm_page_track_register_notifier(struct kvm *kvm,
+int kvm_page_track_register_notifier(struct file *file,
 				     struct kvm_page_track_notifier_node *n);
-void kvm_page_track_unregister_notifier(struct kvm *kvm,
+void kvm_page_track_unregister_notifier(struct file *file,
 					struct kvm_page_track_notifier_node *n);
 
-int kvm_write_track_add_gfn(struct kvm *kvm, gfn_t gfn);
-int kvm_write_track_remove_gfn(struct kvm *kvm, gfn_t gfn);
+int kvm_write_track_add_gfn(struct file *file, gfn_t gfn);
+int kvm_write_track_remove_gfn(struct file *file, gfn_t gfn);
 #else
 /*
  * Allow defining a node in a structure even if page tracking is disabled, e.g.
diff --git a/arch/x86/kvm/Makefile b/arch/x86/kvm/Makefile
index 0474604ab8a1..96f89d6f7e6a 100644
--- a/arch/x86/kvm/Makefile
+++ b/arch/x86/kvm/Makefile
@@ -60,10 +60,7 @@ exports_grep_trailer := --include='*.[ch]' -nrw $(srctree)/virt/kvm $(srctree)/a
 			grep -v -e kvm_page_track_register_notifier \
 				-e kvm_page_track_unregister_notifier \
 				-e kvm_write_track_add_gfn \
-				-e kvm_write_track_remove_gfn \
-				-e kvm_get_kvm \
-				-e kvm_get_kvm_safe \
-				-e kvm_put_kvm
+				-e kvm_write_track_remove_gfn
 
 # Force grep to emit a goofy group separator that can in turn be replaced with
 # the above newline macro (newlines in Make are a nightmare).  Note, grep only
diff --git a/arch/x86/kvm/mmu/page_track.c b/arch/x86/kvm/mmu/page_track.c
index 7e8195a311bb..98f133e4f0b3 100644
--- a/arch/x86/kvm/mmu/page_track.c
+++ b/arch/x86/kvm/mmu/page_track.c
@@ -237,10 +237,11 @@ static int kvm_enable_external_write_tracking(struct kvm *kvm)
  * register the notifier so that event interception for the tracked guest
  * pages can be received.
  */
-int kvm_page_track_register_notifier(struct kvm *kvm,
+int kvm_page_track_register_notifier(struct file *file,
 				     struct kvm_page_track_notifier_node *n)
 {
 	struct kvm_page_track_notifier_head *head;
+	struct kvm *kvm = file_to_kvm(file);
 	int r;
 
 	if (!kvm || kvm->mm != current->mm)
@@ -252,7 +253,7 @@ int kvm_page_track_register_notifier(struct kvm *kvm,
 			return r;
 	}
 
-	kvm_get_kvm(kvm);
+	get_file(file);
 
 	head = &kvm->arch.track_notifier_head;
 
@@ -267,10 +268,11 @@ EXPORT_SYMBOL_GPL(kvm_page_track_register_notifier);
  * stop receiving the event interception. It is the opposed operation of
  * kvm_page_track_register_notifier().
  */
-void kvm_page_track_unregister_notifier(struct kvm *kvm,
+void kvm_page_track_unregister_notifier(struct file *file,
 					struct kvm_page_track_notifier_node *n)
 {
 	struct kvm_page_track_notifier_head *head;
+	struct kvm *kvm = file_to_kvm(file);
 
 	head = &kvm->arch.track_notifier_head;
 
@@ -279,7 +281,7 @@ void kvm_page_track_unregister_notifier(struct kvm *kvm,
 	write_unlock(&kvm->mmu_lock);
 	synchronize_srcu(&head->track_srcu);
 
-	kvm_put_kvm(kvm);
+	fput(file);
 }
 EXPORT_SYMBOL_GPL(kvm_page_track_unregister_notifier);
 
@@ -336,11 +338,12 @@ void kvm_page_track_delete_slot(struct kvm *kvm, struct kvm_memory_slot *slot)
  * add guest page to the tracking pool so that corresponding access on that
  * page will be intercepted.
  *
- * @kvm: the guest instance we are interested in.
+ * @file: the VM file of the guest instance we are interested in.
  * @gfn: the guest page.
  */
-int kvm_write_track_add_gfn(struct kvm *kvm, gfn_t gfn)
+int kvm_write_track_add_gfn(struct file *file, gfn_t gfn)
 {
+	struct kvm *kvm = file_to_kvm(file);
 	struct kvm_memory_slot *slot;
 	int idx;
 
@@ -366,11 +369,12 @@ EXPORT_SYMBOL_GPL(kvm_write_track_add_gfn);
  * remove the guest page from the tracking pool which stops the interception
  * of corresponding access on that page.
  *
- * @kvm: the guest instance we are interested in.
+ * @file: the VM file of the guest instance we are interested in.
  * @gfn: the guest page.
  */
-int kvm_write_track_remove_gfn(struct kvm *kvm, gfn_t gfn)
+int kvm_write_track_remove_gfn(struct file *file, gfn_t gfn)
 {
+	struct kvm *kvm = file_to_kvm(file);
 	struct kvm_memory_slot *slot;
 	int idx;
 
diff --git a/drivers/s390/crypto/vfio_ap_ops.c b/drivers/s390/crypto/vfio_ap_ops.c
index 940c0ff668be..f160e1945fbc 100644
--- a/drivers/s390/crypto/vfio_ap_ops.c
+++ b/drivers/s390/crypto/vfio_ap_ops.c
@@ -1822,17 +1822,27 @@ static const struct attribute_group *vfio_ap_mdev_attr_groups[] = {
 
 /**
  * vfio_ap_mdev_set_kvm - sets all data for @matrix_mdev that are needed
- * to manage AP resources for the guest whose state is represented by @kvm
+ * to manage AP resources for the guest whose state is represented by
+ * @kvm_file
  *
  * @matrix_mdev: a mediated matrix device
- * @kvm: reference to KVM instance
+ * @kvm_file: the KVM VM file this vfio device is associated with
  *
- * Return: 0 if no other mediated matrix device has a reference to @kvm;
+ * Return: 0 if no other mediated matrix device has a reference to the VM;
  * otherwise, returns an -EPERM.
  */
 static int vfio_ap_mdev_set_kvm(struct ap_matrix_mdev *matrix_mdev,
-				struct kvm *kvm)
+				struct file *kvm_file)
 {
+	struct kvm *kvm;
+
+	if (!kvm_file)
+		return -ENOENT;
+
+	kvm = kvm_file->private_data;
+	if (!kvm)
+		return -ENOENT;
+
 	if (kvm->arch.crypto.crycbd) {
 		get_update_locks_for_kvm(kvm);
 		if (kvm->arch.crypto.pqap_hook) {
@@ -1841,7 +1851,6 @@ static int vfio_ap_mdev_set_kvm(struct ap_matrix_mdev *matrix_mdev,
 		}
 		kvm->arch.crypto.pqap_hook = &matrix_mdev->pqap_hook;
 
-		kvm_get_kvm(kvm);
 		matrix_mdev->kvm = kvm;
 		vfio_ap_mdev_update_guest_apcb(matrix_mdev);
 		release_update_locks_for_kvm(kvm);
@@ -1894,7 +1903,6 @@ static void vfio_ap_mdev_unset_kvm(struct ap_matrix_mdev *matrix_mdev)
 		matrix_mdev->kvm = NULL;
 
 		release_update_locks_for_kvm(kvm);
-		kvm_put_kvm(kvm);
 	}
 }
 
diff --git a/drivers/vfio/group.c b/drivers/vfio/group.c
index b2299e5bc6df..5bf8cbdff377 100644
--- a/drivers/vfio/group.c
+++ b/drivers/vfio/group.c
@@ -860,11 +860,20 @@ bool vfio_group_enforced_coherent(struct vfio_group *group)
 	return ret;
 }
 
-void vfio_group_set_kvm(struct vfio_group *group, struct kvm *kvm)
+void vfio_group_set_kvm(struct vfio_group *group, struct file *kvm)
 {
+	struct file *old;
+
+	if (kvm)
+		get_file(kvm);
+
 	spin_lock(&group->kvm_ref_lock);
+	old = group->kvm;
 	group->kvm = kvm;
 	spin_unlock(&group->kvm_ref_lock);
+
+	if (old)
+		fput(old);
 }
 
 /**
diff --git a/drivers/vfio/pci/vfio_pci_zdev.c b/drivers/vfio/pci/vfio_pci_zdev.c
index f47f36314a1c..045132e90bfe 100644
--- a/drivers/vfio/pci/vfio_pci_zdev.c
+++ b/drivers/vfio/pci/vfio_pci_zdev.c
@@ -184,6 +184,7 @@ int vfio_pci_zdev_feature_err(struct vfio_device *device, u32 flags,
 int vfio_pci_zdev_open_device(struct vfio_pci_core_device *vdev)
 {
 	struct zpci_dev *zdev = to_zpci(vdev->pdev);
+	struct kvm *kvm;
 	int ret;
 
 	if (!zdev)
@@ -195,9 +196,14 @@ int vfio_pci_zdev_open_device(struct vfio_pci_core_device *vdev)
 		return 0;
 
 	ret = -ENOENT;
+	kvm = vdev->vdev.kvm->private_data;
+	if (!kvm)
+		goto recovery;
+
 	if (zpci_kvm_hook.kvm_register)
-		ret = zpci_kvm_hook.kvm_register(zdev, vdev->vdev.kvm);
+		ret = zpci_kvm_hook.kvm_register(zdev, kvm);
 
+recovery:
 	if (ret)
 		zpci_stop_mediated_recovery(zdev);
 
diff --git a/drivers/vfio/vfio.h b/drivers/vfio/vfio.h
index 7728bc99b63d..9b619951a5d0 100644
--- a/drivers/vfio/vfio.h
+++ b/drivers/vfio/vfio.h
@@ -23,7 +23,7 @@ struct vfio_device_file {
 	u8 access_granted;
 	u32 devid; /* only valid when iommufd is valid */
 	spinlock_t kvm_ref_lock; /* protect kvm field */
-	struct kvm *kvm;
+	struct file *kvm;
 	struct iommufd_ctx *iommufd; /* protected by struct vfio_device_set::lock */
 };
 
@@ -88,7 +88,7 @@ struct vfio_group {
 #endif
 	enum vfio_group_type		type;
 	struct mutex			group_lock;
-	struct kvm			*kvm;
+	struct file			*kvm;
 	struct file			*opened_file;
 	struct iommufd_ctx		*iommufd;
 	spinlock_t			kvm_ref_lock;
@@ -107,7 +107,7 @@ void vfio_device_group_unuse_iommu(struct vfio_device *device);
 void vfio_df_group_close(struct vfio_device_file *df);
 struct vfio_group *vfio_group_from_file(struct file *file);
 bool vfio_group_enforced_coherent(struct vfio_group *group);
-void vfio_group_set_kvm(struct vfio_group *group, struct kvm *kvm);
+void vfio_group_set_kvm(struct vfio_group *group, struct file *kvm);
 bool vfio_device_has_container(struct vfio_device *device);
 int __init vfio_group_init(void);
 void vfio_group_cleanup(void);
@@ -165,7 +165,7 @@ static inline bool vfio_group_enforced_coherent(struct vfio_group *group)
 	return true;
 }
 
-static inline void vfio_group_set_kvm(struct vfio_group *group, struct kvm *kvm)
+static inline void vfio_group_set_kvm(struct vfio_group *group, struct file *kvm)
 {
 }
 
@@ -429,11 +429,11 @@ static inline void vfio_virqfd_exit(void)
 #endif
 
 #if IS_ENABLED(CONFIG_KVM)
-void vfio_device_get_kvm_safe(struct vfio_device *device, struct kvm *kvm);
+void vfio_device_get_kvm_safe(struct vfio_device *device, struct file *kvm);
 void vfio_device_put_kvm(struct vfio_device *device);
 #else
 static inline void vfio_device_get_kvm_safe(struct vfio_device *device,
-					    struct kvm *kvm)
+					    struct file *kvm)
 {
 }
 
diff --git a/drivers/vfio/vfio_main.c b/drivers/vfio/vfio_main.c
index 423ead48aafe..ed96acfa8635 100644
--- a/drivers/vfio/vfio_main.c
+++ b/drivers/vfio/vfio_main.c
@@ -472,36 +472,14 @@ void vfio_unregister_group_dev(struct vfio_device *device)
 EXPORT_SYMBOL_GPL(vfio_unregister_group_dev);
 
 #if IS_ENABLED(CONFIG_KVM)
-void vfio_device_get_kvm_safe(struct vfio_device *device, struct kvm *kvm)
+void vfio_device_get_kvm_safe(struct vfio_device *device, struct file *kvm)
 {
-	void (*pfn)(struct kvm *kvm);
-	bool (*fn)(struct kvm *kvm);
-	bool ret;
-
 	lockdep_assert_held(&device->dev_set->lock);
 
 	if (!kvm)
 		return;
 
-	pfn = symbol_get(kvm_put_kvm);
-	if (WARN_ON(!pfn))
-		return;
-
-	fn = symbol_get(kvm_get_kvm_safe);
-	if (WARN_ON(!fn)) {
-		symbol_put(kvm_put_kvm);
-		return;
-	}
-
-	ret = fn(kvm);
-	symbol_put(kvm_get_kvm_safe);
-	if (!ret) {
-		symbol_put(kvm_put_kvm);
-		return;
-	}
-
-	device->put_kvm = pfn;
-	device->kvm = kvm;
+	device->kvm = get_file(kvm);
 }
 
 void vfio_device_put_kvm(struct vfio_device *device)
@@ -511,14 +489,7 @@ void vfio_device_put_kvm(struct vfio_device *device)
 	if (!device->kvm)
 		return;
 
-	if (WARN_ON(!device->put_kvm))
-		goto clear;
-
-	device->put_kvm(device->kvm);
-	device->put_kvm = NULL;
-	symbol_put(kvm_put_kvm);
-
-clear:
+	fput(device->kvm);
 	device->kvm = NULL;
 }
 #endif
@@ -1544,9 +1515,13 @@ bool vfio_file_enforced_coherent(struct file *file)
 }
 EXPORT_SYMBOL_GPL(vfio_file_enforced_coherent);
 
-static void vfio_device_file_set_kvm(struct file *file, struct kvm *kvm)
+static void vfio_device_file_set_kvm(struct file *file, struct file *kvm)
 {
 	struct vfio_device_file *df = file->private_data;
+	struct file *old;
+
+	if (kvm)
+		get_file(kvm);
 
 	/*
 	 * The kvm is first recorded in the vfio_device_file, and will
@@ -1554,28 +1529,32 @@ static void vfio_device_file_set_kvm(struct file *file, struct kvm *kvm)
 	 * iommufd successfully in the vfio device cdev path.
 	 */
 	spin_lock(&df->kvm_ref_lock);
+	old = df->kvm;
 	df->kvm = kvm;
 	spin_unlock(&df->kvm_ref_lock);
+
+	if (old)
+		fput(old);
 }
 
 /**
  * vfio_file_set_kvm - Link a kvm with VFIO drivers
- * @file: VFIO group file or VFIO device file
- * @kvm: KVM to link
+ * @vfio_file: VFIO group file or VFIO device file
+ * @kvm: KVM file to link
  *
  * When a VFIO device is first opened the KVM will be available in
  * device->kvm if one was associated with the file.
  */
-void vfio_file_set_kvm(struct file *file, struct kvm *kvm)
+void vfio_file_set_kvm(struct file *vfio_file, struct file *kvm)
 {
 	struct vfio_group *group;
 
-	group = vfio_group_from_file(file);
+	group = vfio_group_from_file(vfio_file);
 	if (group)
 		vfio_group_set_kvm(group, kvm);
 
-	if (vfio_device_from_file(file))
-		vfio_device_file_set_kvm(file, kvm);
+	if (vfio_device_from_file(vfio_file))
+		vfio_device_file_set_kvm(vfio_file, kvm);
 }
 EXPORT_SYMBOL_GPL(vfio_file_set_kvm);
 
diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
index 03bfc92864b6..bfecce1b5955 100644
--- a/include/linux/kvm_host.h
+++ b/include/linux/kvm_host.h
@@ -784,6 +784,13 @@ struct kvm {
 	 * kvm_swap_active_memslots().
 	 */
 	struct mutex slots_arch_lock;
+	/*
+	 * Back-reference to the VM file for subsystems (e.g., VFIO). Holds no
+	 * reference to avoid pinning the VM. Access via get_file_active(&kvm->file)
+	 * only—safe because cleared in kvm_vm_release() before file freed. Callers
+	 * storing the result must take their own reference via get_file()
+	 */
+	struct file *file;
 	struct mm_struct *mm; /* userspace tied to this vm */
 	unsigned long nr_memslot_pages;
 	/* The two memslot sets - active and inactive (per address space) */
@@ -1082,6 +1089,7 @@ void kvm_get_kvm(struct kvm *kvm);
 bool kvm_get_kvm_safe(struct kvm *kvm);
 void kvm_put_kvm(struct kvm *kvm);
 bool file_is_kvm(struct file *file);
+struct kvm *file_to_kvm(struct file *file);
 void kvm_put_kvm_no_destroy(struct kvm *kvm);
 
 static inline struct kvm_memslots *__kvm_memslots(struct kvm *kvm, int as_id)
diff --git a/include/linux/vfio.h b/include/linux/vfio.h
index 45f08986359e..0cc91c6f96d2 100644
--- a/include/linux/vfio.h
+++ b/include/linux/vfio.h
@@ -54,7 +54,7 @@ struct vfio_device {
 	struct list_head dev_set_list;
 	unsigned int migration_flags;
 	u8 precopy_info_v2;
-	struct kvm *kvm;
+	struct file *kvm;
 
 	/* Members below here are private, not for driver use */
 	unsigned int index;
@@ -66,7 +66,6 @@ struct vfio_device {
 	unsigned int open_count;
 	struct completion comp;
 	struct iommufd_access *iommufd_access;
-	void (*put_kvm)(struct kvm *kvm);
 	struct inode *inode;
 #if IS_ENABLED(CONFIG_IOMMUFD)
 	struct iommufd_device *iommufd_device;
@@ -378,7 +377,7 @@ static inline bool vfio_file_has_dev(struct file *file, struct vfio_device *devi
 #endif
 bool vfio_file_is_valid(struct file *file);
 bool vfio_file_enforced_coherent(struct file *file);
-void vfio_file_set_kvm(struct file *file, struct kvm *kvm);
+void vfio_file_set_kvm(struct file *vfio_file, struct file *kvm);
 
 #define VFIO_PIN_PAGES_MAX_ENTRIES	(PAGE_SIZE/sizeof(unsigned long))
 
diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index 65eb26a0520d..34b4c43908b3 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -1314,7 +1314,7 @@ void kvm_get_kvm(struct kvm *kvm)
 {
 	refcount_inc(&kvm->users_count);
 }
-EXPORT_SYMBOL_GPL(kvm_get_kvm);
+EXPORT_SYMBOL_FOR_KVM_INTERNAL(kvm_get_kvm);
 
 /*
  * Make sure the vm is not during destruction, which is a safe version of
@@ -1324,14 +1324,14 @@ bool kvm_get_kvm_safe(struct kvm *kvm)
 {
 	return refcount_inc_not_zero(&kvm->users_count);
 }
-EXPORT_SYMBOL_GPL(kvm_get_kvm_safe);
+EXPORT_SYMBOL_FOR_KVM_INTERNAL(kvm_get_kvm_safe);
 
 void kvm_put_kvm(struct kvm *kvm)
 {
 	if (refcount_dec_and_test(&kvm->users_count))
 		kvm_destroy_vm(kvm);
 }
-EXPORT_SYMBOL_GPL(kvm_put_kvm);
+EXPORT_SYMBOL_FOR_KVM_INTERNAL(kvm_put_kvm);
 
 /*
  * Used to put a reference that was taken on behalf of an object associated
@@ -1352,6 +1352,8 @@ static int kvm_vm_release(struct inode *inode, struct file *filp)
 
 	kvm_irqfd_release(kvm);
 
+	WRITE_ONCE(kvm->file, NULL);
+
 	kvm_put_kvm(kvm);
 	return 0;
 }
@@ -5496,6 +5498,15 @@ bool file_is_kvm(struct file *file)
 }
 EXPORT_SYMBOL_FOR_KVM_INTERNAL(file_is_kvm);
 
+struct kvm *file_to_kvm(struct file *file)
+{
+	if (!file_is_kvm(file))
+		return NULL;
+
+	return file->private_data;
+}
+EXPORT_SYMBOL_FOR_KVM_INTERNAL(file_to_kvm);
+
 static int kvm_dev_ioctl_create_vm(unsigned long type)
 {
 	char fdname[ITOA_MAX_LEN + 1];
@@ -5527,6 +5538,9 @@ static int kvm_dev_ioctl_create_vm(unsigned long type)
 	 * cases it will be called by the final fput(file) and will take
 	 * care of doing kvm_put_kvm(kvm).
 	 */
+
+	kvm->file = file;
+
 	kvm_uevent_notify_change(KVM_EVENT_CREATE_VM, kvm);
 
 	fd_install(fd, file);
diff --git a/virt/kvm/vfio.c b/virt/kvm/vfio.c
index 6cdc4e9a333a..19548a430942 100644
--- a/virt/kvm/vfio.c
+++ b/virt/kvm/vfio.c
@@ -35,15 +35,15 @@ struct kvm_vfio {
 	bool noncoherent;
 };
 
-static void kvm_vfio_file_set_kvm(struct file *file, struct kvm *kvm)
+static void kvm_vfio_file_set_kvm(struct file *vfio_file, struct file *kvm)
 {
-	void (*fn)(struct file *file, struct kvm *kvm);
+	void (*fn)(struct file *vfio_file, struct file *kvm);
 
 	fn = symbol_get(vfio_file_set_kvm);
 	if (!fn)
 		return;
 
-	fn(file, kvm);
+	fn(vfio_file, kvm);
 
 	symbol_put(vfio_file_set_kvm);
 }
@@ -144,6 +144,7 @@ static int kvm_vfio_file_add(struct kvm_device *dev, unsigned int fd)
 {
 	struct kvm_vfio *kv = dev->private;
 	struct kvm_vfio_file *kvf;
+	struct file *kvm_file __free(fput) = NULL;
 	struct file *filp __free(fput) = NULL;
 
 	filp = fget(fd);
@@ -154,6 +155,10 @@ static int kvm_vfio_file_add(struct kvm_device *dev, unsigned int fd)
 	if (!kvm_vfio_file_is_valid(filp))
 		return -EINVAL;
 
+	kvm_file = get_file_active(&dev->kvm->file);
+	if (!kvm_file)
+		return -ENOENT;
+
 	guard(mutex)(&kv->lock);
 
 	list_for_each_entry(kvf, &kv->file_list, node) {
@@ -168,7 +173,7 @@ static int kvm_vfio_file_add(struct kvm_device *dev, unsigned int fd)
 	kvf->file = get_file(filp);
 	list_add_tail(&kvf->node, &kv->file_list);
 
-	kvm_vfio_file_set_kvm(kvf->file, dev->kvm);
+	kvm_vfio_file_set_kvm(kvf->file, kvm_file);
 	kvm_vfio_update_coherency(dev);
 
 	return 0;

---
base-commit: 940de590b839f71d6dc846160534bf202401b8b7
change-id: 20260812-vfio-f9069d8ab545

Best regards,
-- 
Steffen Eiden <seiden@linux.ibm.com>

Re: [PATCH v2] vfio: Use file-based reference counting for KVM

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:

New issues:
- [Critical] Untyped `struct file *` for KVM VM leads to type confusion and out-of-bounds access in VFIO device drivers when multiple concurrent KVM modules are loaded.

Pre-existing issues:
- [High] Pre-existing bug in the cdev IOMMUFD binding path allows concurrent opens to leak the KVM file reference and prematurely clear `device->kvm`.
--

commit dd4fd68a7d7d9888bc552f388401d95d1f137c5f
Author: Steffen Eiden <seiden@linux.ibm.com>

vfio: Use file-based reference counting for KVM

This patch replaces manual module reference counting with file-based reference
counting for KVM integration in VFIO. It passes struct file pointers instead
of struct kvm pointers across the interface to leverage the kernel's existing
file reference counting mechanism. This removes exported symbols and supports
multiple concurrent KVM modules.

Link: https://lore.kernel.org/all/20260831144802.834315-1-seiden@linux.ibm.com/
Link: https://lore.kernel.org/r/20260812-vfio-v1-1-5cfe0b1fa4e7@linux.ibm.com

> diff --git a/drivers/s390/crypto/vfio_ap_ops.c b/drivers/s390/crypto/vfio_ap_ops.c
> index 940c0ff668bee..f160e1945fbce 100644
> --- a/drivers/s390/crypto/vfio_ap_ops.c
> +++ b/drivers/s390/crypto/vfio_ap_ops.c

[ ... ]

> @@ -1822,17 +1822,27 @@ static const struct attribute_group *vfio_ap_mdev_attr_groups[] = {
>  
>  /**
>   * vfio_ap_mdev_set_kvm - sets all data for @matrix_mdev that are needed
> - * to manage AP resources for the guest whose state is represented by @kvm
> + * to manage AP resources for the guest whose state is represented by
> + * @kvm_file
>   *
>   * @matrix_mdev: a mediated matrix device
> - * @kvm: reference to KVM instance
> + * @kvm_file: the KVM VM file this vfio device is associated with
>   *
> - * Return: 0 if no other mediated matrix device has a reference to @kvm;
> + * Return: 0 if no other mediated matrix device has a reference to the VM;
>   * otherwise, returns an -EPERM.
>   */
>  static int vfio_ap_mdev_set_kvm(struct ap_matrix_mdev *matrix_mdev,
> -				struct kvm *kvm)
> +				struct file *kvm_file)
>  {
> +	struct kvm *kvm;
> +
> +	if (!kvm_file)
> +		return -ENOENT;
> +
> +	kvm = kvm_file->private_data;
> +	if (!kvm)
> +		return -ENOENT;
> +
>  	if (kvm->arch.crypto.crycbd) {

[Severity: Critical]
Does this code adequately verify that the file is an s390 KVM file before
casting its private_data in vfio_ap_mdev_set_kvm() inside
drivers/s390/crypto/vfio_ap_ops.c?

If a user instantiates a VM using a concurrent KVM-arm64 module and attaches
an s390-specific VFIO device to its KVM device, could this driver extract the
untyped private_data and mistakenly treat an arm64 struct kvm as an s390 one?

Because the memory layouts of struct kvm_arch differ across architectures,
would accessing s390-specific fields in the arch member (like
kvm->arch.crypto.pqap_hook) result in arbitrary memory corruption?

>  		get_update_locks_for_kvm(kvm);
>  		if (kvm->arch.crypto.pqap_hook) {

[ ... ]

> diff --git a/drivers/vfio/pci/vfio_pci_zdev.c b/drivers/vfio/pci/vfio_pci_zdev.c
> index f47f36314a1cc..045132e90bfee 100644
> --- a/drivers/vfio/pci/vfio_pci_zdev.c
> +++ b/drivers/vfio/pci/vfio_pci_zdev.c
> @@ -184,6 +184,7 @@ int vfio_pci_zdev_feature_err(struct vfio_device *device, u32 flags,
>  int vfio_pci_zdev_open_device(struct vfio_pci_core_device *vdev)
>  {
>  	struct zpci_dev *zdev = to_zpci(vdev->pdev);
> +	struct kvm *kvm;
>  	int ret;
>  
>  	if (!zdev)
> @@ -195,9 +196,14 @@ int vfio_pci_zdev_open_device(struct vfio_pci_core_device *vdev)
>  		return 0;
>  
>  	ret = -ENOENT;
> +	kvm = vdev->vdev.kvm->private_data;
> +	if (!kvm)
> +		goto recovery;
> +
>  	if (zpci_kvm_hook.kvm_register)
> -		ret = zpci_kvm_hook.kvm_register(zdev, vdev->vdev.kvm);
> +		ret = zpci_kvm_hook.kvm_register(zdev, kvm);

[Severity: Critical]
Is it possible for a similar regression to occur here in
vfio_pci_zdev_open_device() within drivers/vfio/pci/vfio_pci_zdev.c?

If an arm64 VM file is passed via vdev->vdev.kvm, passing it to
zpci_kvm_hook.kvm_register() could potentially cause type confusion in the
host kernel.

>  
> +recovery:
>  	if (ret)
>  		zpci_stop_mediated_recovery(zdev);

[ ... ]

> diff --git a/drivers/vfio/vfio_main.c b/drivers/vfio/vfio_main.c
> index 423ead48aafe3..ed96acfa86358 100644
> --- a/drivers/vfio/vfio_main.c
> +++ b/drivers/vfio/vfio_main.c

[ ... ]

> @@ -511,14 +489,7 @@ void vfio_device_put_kvm(struct vfio_device *device)
>  	if (!device->kvm)
>  		return;
>  
> -	if (WARN_ON(!device->put_kvm))
> -		goto clear;
> -
> -	device->put_kvm(device->kvm);
> -	device->put_kvm = NULL;
> -	symbol_put(kvm_put_kvm);
> -
> -clear:
> +	fput(device->kvm);
>  	device->kvm = NULL;

[Severity: High]
This is a pre-existing issue, but does this error path leave the first active
file descriptor with a NULL KVM pointer during concurrent cdev IOMMUFD binds?

Looking at vfio_df_ioctl_bind_iommufd() in drivers/vfio/device_cdev.c, if a
user opens a VFIO cdev twice, obtaining two vfio_device_file (df) instances:

1. bind_iommufd on the first df succeeds, setting device->kvm and
device->open_count.

2. bind_iommufd on the second df bypasses the df->access_granted check and
unconditionally calls vfio_df_get_kvm_safe(df). If a KVM file was linked to
the second df, this overwrites device->kvm with the second file's reference
without releasing the first, causing a resource leak.

3. The subsequent vfio_df_open(df) fails because device->open_count != 0.

4. The error path then calls vfio_device_put_kvm() in
drivers/vfio/vfio_main.c, which drops the reference and sets
device->kvm = NULL.

Could this sequence leave the first df operating with a NULL device->kvm,
leading to potential NULL pointer dereferences in drivers?

>  }
>  #endif

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260903-vfio-v2-1-ef4cd4190ae7@linux.ibm.com?part=1