kvm/vfio: ensure kvg instance stays around in kvm_vfio_group_add()
authorDmitry Torokhov <dmitry.torokhov@gmail.com>
Fri, 14 Jul 2023 22:45:32 +0000 (15:45 -0700)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Wed, 13 Sep 2023 07:42:46 +0000 (09:42 +0200)
commita4ff4b54f38825a9cdbdd1d70d2341af99ff67e6
tree89dcb8b7ec3926e6bc299e834d491853ab0fbcca
parentfef33ca5e28c876ddeddd586d090fd872a06fca7
kvm/vfio: ensure kvg instance stays around in kvm_vfio_group_add()

[ Upstream commit 9e0f4f2918c2ff145d3dedee862d9919a6ed5812 ]

kvm_vfio_group_add() creates kvg instance, links it to kv->group_list,
and calls kvm_vfio_file_set_kvm() with kvg->file as an argument after
dropping kv->lock. If we race group addition and deletion calls, kvg
instance may get freed by the time we get around to calling
kvm_vfio_file_set_kvm().

Previous iterations of the code did not reference kvg->file outside of
the critical section, but used a temporary variable. Still, they had
similar problem of the file reference being owned by kvg structure and
potential for kvm_vfio_group_del() dropping it before
kvm_vfio_group_add() had a chance to complete.

Fix this by moving call to kvm_vfio_file_set_kvm() under the protection
of kv->lock. We already call it while holding the same lock when vfio
group is being deleted, so it should be safe here as well.

Fixes: 2fc1bec15883 ("kvm: set/clear kvm to/from vfio_group when group add/delete")
Reviewed-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Reviewed-by: Kevin Tian <kevin.tian@intel.com>
Link: https://lore.kernel.org/r/20230714224538.404793-1-dmitry.torokhov@gmail.com
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
virt/kvm/vfio.c