[v3,1/2] kvm/vfio: ensure kvg instance stays around in kvm_vfio_group_add()

  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>
---

v3: added Alex's reviewed-by

v2: updated commit description with the correct "Fixes" tag (per Alex),
    expanded commit description to mention issues with the earlier
    implementation of kvm_vfio_group_add().

 virt/kvm/vfio.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
  

> From: Dmitry Torokhov <dmitry.torokhov@gmail.com>
> Sent: Saturday, July 15, 2023 6:46 AM
> 
> 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>
  

Hey Paolo,

I'll pull this through the vfio tree unless you have a particular
interest.  Thanks,

Alex

On Fri, 14 Jul 2023 15:45:32 -0700
Dmitry Torokhov <dmitry.torokhov@gmail.com> wrote:

> 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>
> ---
> 
> v3: added Alex's reviewed-by
> 
> v2: updated commit description with the correct "Fixes" tag (per Alex),
>     expanded commit description to mention issues with the earlier
>     implementation of kvm_vfio_group_add().
> 
>  virt/kvm/vfio.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/virt/kvm/vfio.c b/virt/kvm/vfio.c
> index 9584eb57e0ed..cd46d7ef98d6 100644
> --- a/virt/kvm/vfio.c
> +++ b/virt/kvm/vfio.c
> @@ -179,10 +179,10 @@ static int kvm_vfio_group_add(struct kvm_device *dev, unsigned int fd)
>  	list_add_tail(&kvg->node, &kv->group_list);
>  
>  	kvm_arch_start_assignment(dev->kvm);
> +	kvm_vfio_file_set_kvm(kvg->file, dev->kvm);
>  
>  	mutex_unlock(&kv->lock);
>  
> -	kvm_vfio_file_set_kvm(kvg->file, dev->kvm);
>  	kvm_vfio_update_coherency(dev);
>  
>  	return 0;
  

On Fri, 14 Jul 2023 15:45:32 -0700
Dmitry Torokhov <dmitry.torokhov@gmail.com> wrote:

> 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>
> ---

Applied series to vfio next branch for v6.6.  There's a minor rebase
involved, so please double check the results:

https://github.com/awilliam/linux-vfio/commits/next

Thanks,
Alex

> 
> v3: added Alex's reviewed-by
> 
> v2: updated commit description with the correct "Fixes" tag (per Alex),
>     expanded commit description to mention issues with the earlier
>     implementation of kvm_vfio_group_add().
> 
>  virt/kvm/vfio.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/virt/kvm/vfio.c b/virt/kvm/vfio.c
> index 9584eb57e0ed..cd46d7ef98d6 100644
> --- a/virt/kvm/vfio.c
> +++ b/virt/kvm/vfio.c
> @@ -179,10 +179,10 @@ static int kvm_vfio_group_add(struct kvm_device *dev, unsigned int fd)
>  	list_add_tail(&kvg->node, &kv->group_list);
>  
>  	kvm_arch_start_assignment(dev->kvm);
> +	kvm_vfio_file_set_kvm(kvg->file, dev->kvm);
>  
>  	mutex_unlock(&kv->lock);
>  
> -	kvm_vfio_file_set_kvm(kvg->file, dev->kvm);
>  	kvm_vfio_update_coherency(dev);
>  
>  	return 0;
  

On Thu, Aug 03, 2023 at 01:38:12PM -0600, Alex Williamson wrote:
> On Fri, 14 Jul 2023 15:45:32 -0700
> Dmitry Torokhov <dmitry.torokhov@gmail.com> wrote:
> 
> > 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>
> > ---
> 
> Applied series to vfio next branch for v6.6.  There's a minor rebase
> involved, so please double check the results:
> 
> https://github.com/awilliam/linux-vfio/commits/next

Looks good to me, thanks!