Re: [PATCH 2/4] drm: Add drm_object lease infrastructure

2017-04-09 Thread Michel Dänzer 
On 03/04/17 01:31 AM, Keith Packard wrote:
> Daniel Vetter  writes:
> 
>> I'm also not sure whether we really want sub-leases in v1, that's easy
>> to add later on, but for now just complicates stuff. Main compositor
>> should be a full master, VR can be the first lease level, we don't
>> need more I think for now?
> 
> We've discussed how leases might be used to implement multi-user
> support, so offering sub-leases means that environment could also
> support leasing resources out from the users session.
> 
> We also just don't know how useful it might be until we explore the
> space a bit more.

It should only be added upstream once it's clear that it's useful.


> Given that it takes years to get new features into distributions, I
> tend to error on the side of generality.

Why would it take years? Distros seem to generally use the latest
upstream kernel release available at release/freeze.


-- 
Earthling Michel Dänzer   |   http://www.amd.com
Libre software enthusiast | Mesa and X developer



signature.asc
Description: OpenPGP digital signature

Re: [PATCH 2/4] drm: Add drm_object lease infrastructure

2017-04-09 Thread Michel Dänzer 
On 03/04/17 01:31 AM, Keith Packard wrote:
> Daniel Vetter  writes:
> 
>> I'm also not sure whether we really want sub-leases in v1, that's easy
>> to add later on, but for now just complicates stuff. Main compositor
>> should be a full master, VR can be the first lease level, we don't
>> need more I think for now?
> 
> We've discussed how leases might be used to implement multi-user
> support, so offering sub-leases means that environment could also
> support leasing resources out from the users session.
> 
> We also just don't know how useful it might be until we explore the
> space a bit more.

It should only be added upstream once it's clear that it's useful.


> Given that it takes years to get new features into distributions, I
> tend to error on the side of generality.

Why would it take years? Distros seem to generally use the latest
upstream kernel release available at release/freeze.


-- 
Earthling Michel Dänzer   |   http://www.amd.com
Libre software enthusiast | Mesa and X developer



signature.asc
Description: OpenPGP digital signature

Re: [PATCH 2/4] drm: Add drm_object lease infrastructure

2017-04-02 Thread Keith Packard 
Daniel Vetter  writes:

> On Sat, Apr 01, 2017 at 10:08:39AM -0700, Keith Packard wrote:
>> +BUG_ON(__mutex_owner(>dev->mode_config.idr_mutex) != current);
>
> Forgot to reply on this:
>
> lockdep_assert_held + enable lockdep.

Thanks. Will fix.

-- 
-keith


signature.asc
Description: PGP signature

Re: [PATCH 2/4] drm: Add drm_object lease infrastructure

2017-04-02 Thread Keith Packard 
Daniel Vetter  writes:

> On Sat, Apr 01, 2017 at 10:08:39AM -0700, Keith Packard wrote:
>> +BUG_ON(__mutex_owner(>dev->mode_config.idr_mutex) != current);
>
> Forgot to reply on this:
>
> lockdep_assert_held + enable lockdep.

Thanks. Will fix.

-- 
-keith


signature.asc
Description: PGP signature

Re: [PATCH 2/4] drm: Add drm_object lease infrastructure

2017-04-02 Thread Daniel Vetter 
On Sat, Apr 01, 2017 at 10:08:39AM -0700, Keith Packard wrote:
> + BUG_ON(__mutex_owner(>dev->mode_config.idr_mutex) != current);

Forgot to reply on this:

lockdep_assert_held + enable lockdep.

Cheers, Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

Re: [PATCH 2/4] drm: Add drm_object lease infrastructure

2017-04-02 Thread Daniel Vetter 
On Sat, Apr 01, 2017 at 10:08:39AM -0700, Keith Packard wrote:
> + BUG_ON(__mutex_owner(>dev->mode_config.idr_mutex) != current);

Forgot to reply on this:

lockdep_assert_held + enable lockdep.

Cheers, Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

Re: [PATCH 2/4] drm: Add drm_object lease infrastructure

2017-04-02 Thread Keith Packard 
Daniel Vetter  writes:

> Still not sure we want to restrict objects on the lessor side. Feels like
> unecessary complexity (i.e. more bugs in kernel, that's never good), and
> at best only needed for lessors who can't keep track of stuff.

It's been useful when hacking existing code, and will help catch
application bugs. Limiting access to what you actually need always seems
like good practice to me.

> I'm also not sure whether we really want sub-leases in v1, that's easy
> to add later on, but for now just complicates stuff. Main compositor
> should be a full master, VR can be the first lease level, we don't
> need more I think for now?

We've discussed how leases might be used to implement multi-user
support, so offering sub-leases means that environment could also
support leasing resources out from the users session.

We also just don't know how useful it might be until we explore the
space a bit more. Given that it takes years to get new features into
distributions, I tend to error on the side of generality.

I think a key requirement for acceptance would be a set of robust tests,
something I haven't started writing yet.

>> +/* Tree of display resource leases, each of which is a drm_master struct
>> + * All of these get activated simultaneously, so drm_device master 
>> points
>
> _device.master to do a reference in kernel-doc. Please feed this to
> kernel-doc in general and make sure the links all point at the right
> stuff, and it's all parsed.

Thanks, will do.

-- 
-keith


signature.asc
Description: PGP signature

Re: [PATCH 2/4] drm: Add drm_object lease infrastructure

2017-04-02 Thread Keith Packard 
Daniel Vetter  writes:

> Still not sure we want to restrict objects on the lessor side. Feels like
> unecessary complexity (i.e. more bugs in kernel, that's never good), and
> at best only needed for lessors who can't keep track of stuff.

It's been useful when hacking existing code, and will help catch
application bugs. Limiting access to what you actually need always seems
like good practice to me.

> I'm also not sure whether we really want sub-leases in v1, that's easy
> to add later on, but for now just complicates stuff. Main compositor
> should be a full master, VR can be the first lease level, we don't
> need more I think for now?

We've discussed how leases might be used to implement multi-user
support, so offering sub-leases means that environment could also
support leasing resources out from the users session.

We also just don't know how useful it might be until we explore the
space a bit more. Given that it takes years to get new features into
distributions, I tend to error on the side of generality.

I think a key requirement for acceptance would be a set of robust tests,
something I haven't started writing yet.

>> +/* Tree of display resource leases, each of which is a drm_master struct
>> + * All of these get activated simultaneously, so drm_device master 
>> points
>
> _device.master to do a reference in kernel-doc. Please feed this to
> kernel-doc in general and make sure the links all point at the right
> stuff, and it's all parsed.

Thanks, will do.

-- 
-keith


signature.asc
Description: PGP signature

Re: [PATCH 2/4] drm: Add drm_object lease infrastructure

2017-04-02 Thread Daniel Vetter 
On Sat, Apr 01, 2017 at 10:08:39AM -0700, Keith Packard wrote:
> This provides new data structures to hold "lease" information about
> drm mode setting objects, and provides for creating new drm_masters
> which have access to a subset of the available drm resources.
> 
> An 'owner' is a drm_master which is not leasing the objects from
> another drm_master, and hence 'owns' them. This sits at the top of a
> tree of drm_masters.
> 
> A 'lessee' is a drm_master which is leasing objects from some other
> drm_master. Each lessee holds the set of objects which it is leasing
> from the lessor.
> 
> A 'lessor' is a drm_master which is leasing objects to another
> drm_master.
> 
> The set of objects any drm_master 'controls' is limited to the set of
> objects it leases (for lessees) or all objects (for owners),
> optionally minus the set of objects it has leased to other
> drm_masters.

Still not sure we want to restrict objects on the lessor side. Feels like
unecessary complexity (i.e. more bugs in kernel, that's never good), and
at best only needed for lessors who can't keep track of stuff. I'm also
not sure whether we really want sub-leases in v1, that's easy to add later
on, but for now just complicates stuff. Main compositor should be a full
master, VR can be the first lease level, we don't need more I think for
now?

> Objects not controlled by a drm_master cannot be modified through the
> various state manipulating ioctls, and any state reported back to user
> space will be edited to make them appear idle and/or unusable. For
> instance, connectors always report 'disconnected', while encoders
> report no possible crtcs or clones.
> 
> The full list of lessees leasing objects from an owner (either
> directly, or indirectly through another lessee), can be searched from
> an idr in the drm_master of the owner.
> 
> Signed-off-by: Keith Packard 

[snip]

> diff --git a/include/drm/drm_auth.h b/include/drm/drm_auth.h
> index 610223b0481b..e0e2af09d3af 100644
> --- a/include/drm/drm_auth.h
> +++ b/include/drm/drm_auth.h
> @@ -50,10 +50,38 @@ struct drm_master {
>   struct idr magic_map;
>   struct drm_lock_data lock;
>   void *driver_priv;
> +
> + /* Tree of display resource leases, each of which is a drm_master struct
> +  * All of these get activated simultaneously, so drm_device master 
> points

_device.master to do a reference in kernel-doc. Please feed this to
kernel-doc in general and make sure the links all point at the right
stuff, and it's all parsed.

Cheers, Daniel


> +  * at the top of the tree (for which lessor is NULL)
> +  */
> +
> + /** Lease holder */

/** @lessor: Lease holder. */

> + struct drm_master *lessor;
> +
> + /** id for lessees. Owners always have id 0 */
> + int lessee_id;
> +
> + /** other lessees of the same master */
> + struct list_head lessee_list;
> +
> + /** drm_masters leasing from this one */
> + struct list_head lessees;
> +
> + /** Objects leased to this drm_master. */
> + struct idr leases;
> +
> + /** All lessees under this owner (only used where lessor == NULL) */
> + struct idr lessee_idr;
> +
> + /** Indicates that the leased objects should be hidden from the lessor 
> */
> + bool mask_lease;
>  };
>  
>  struct drm_master *drm_master_get(struct drm_master *master);
>  void drm_master_put(struct drm_master **master);
>  bool drm_is_current_master(struct drm_file *fpriv);
>  
> +struct drm_master *drm_master_create(struct drm_device *dev);
> +
>  #endif
> diff --git a/include/drm/drm_lease.h b/include/drm/drm_lease.h
> new file mode 100644
> index ..e02adf3e42fd
> --- /dev/null
> +++ b/include/drm/drm_lease.h
> @@ -0,0 +1,51 @@
> +/*
> + * Copyright © 2017 Keith Packard 
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License as published by
> + * the Free Software Foundation, either version 2 of the License, or
> + * (at your option) any later version.
> + *
> + * This program is distributed in the hope that it will be useful, but
> + * WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> + * General Public License for more details.
> + */
> +
> +#ifndef _DRM_LEASE_H_
> +#define _DRM_LEASE_H_
> +
> +struct drm_file;
> +struct drm_device;
> +
> +struct drm_master *drm_lease_owner(struct drm_master *master);
> +
> +struct drm_master *drm_lessee_find(struct drm_master *top, int lessee_id);
> +
> +void drm_lease_destroy(struct drm_master *lessee);
> +
> +struct drm_mode_object *drm_lease_find(struct drm_master *master, int id);
> +
> +/**
> + * drm_lease_check - check drm_mode_object lease status
> + * @master: the drm_master
> + * @id: the object id
> + *
> + * Checks if the specified master holds a lease on the object
> + * and also whether it has been leased to

Re: [PATCH 2/4] drm: Add drm_object lease infrastructure

2017-04-02 Thread Daniel Vetter 
On Sat, Apr 01, 2017 at 10:08:39AM -0700, Keith Packard wrote:
> This provides new data structures to hold "lease" information about
> drm mode setting objects, and provides for creating new drm_masters
> which have access to a subset of the available drm resources.
> 
> An 'owner' is a drm_master which is not leasing the objects from
> another drm_master, and hence 'owns' them. This sits at the top of a
> tree of drm_masters.
> 
> A 'lessee' is a drm_master which is leasing objects from some other
> drm_master. Each lessee holds the set of objects which it is leasing
> from the lessor.
> 
> A 'lessor' is a drm_master which is leasing objects to another
> drm_master.
> 
> The set of objects any drm_master 'controls' is limited to the set of
> objects it leases (for lessees) or all objects (for owners),
> optionally minus the set of objects it has leased to other
> drm_masters.

Still not sure we want to restrict objects on the lessor side. Feels like
unecessary complexity (i.e. more bugs in kernel, that's never good), and
at best only needed for lessors who can't keep track of stuff. I'm also
not sure whether we really want sub-leases in v1, that's easy to add later
on, but for now just complicates stuff. Main compositor should be a full
master, VR can be the first lease level, we don't need more I think for
now?

> Objects not controlled by a drm_master cannot be modified through the
> various state manipulating ioctls, and any state reported back to user
> space will be edited to make them appear idle and/or unusable. For
> instance, connectors always report 'disconnected', while encoders
> report no possible crtcs or clones.
> 
> The full list of lessees leasing objects from an owner (either
> directly, or indirectly through another lessee), can be searched from
> an idr in the drm_master of the owner.
> 
> Signed-off-by: Keith Packard 

[snip]

> diff --git a/include/drm/drm_auth.h b/include/drm/drm_auth.h
> index 610223b0481b..e0e2af09d3af 100644
> --- a/include/drm/drm_auth.h
> +++ b/include/drm/drm_auth.h
> @@ -50,10 +50,38 @@ struct drm_master {
>   struct idr magic_map;
>   struct drm_lock_data lock;
>   void *driver_priv;
> +
> + /* Tree of display resource leases, each of which is a drm_master struct
> +  * All of these get activated simultaneously, so drm_device master 
> points

_device.master to do a reference in kernel-doc. Please feed this to
kernel-doc in general and make sure the links all point at the right
stuff, and it's all parsed.

Cheers, Daniel


> +  * at the top of the tree (for which lessor is NULL)
> +  */
> +
> + /** Lease holder */

/** @lessor: Lease holder. */

> + struct drm_master *lessor;
> +
> + /** id for lessees. Owners always have id 0 */
> + int lessee_id;
> +
> + /** other lessees of the same master */
> + struct list_head lessee_list;
> +
> + /** drm_masters leasing from this one */
> + struct list_head lessees;
> +
> + /** Objects leased to this drm_master. */
> + struct idr leases;
> +
> + /** All lessees under this owner (only used where lessor == NULL) */
> + struct idr lessee_idr;
> +
> + /** Indicates that the leased objects should be hidden from the lessor 
> */
> + bool mask_lease;
>  };
>  
>  struct drm_master *drm_master_get(struct drm_master *master);
>  void drm_master_put(struct drm_master **master);
>  bool drm_is_current_master(struct drm_file *fpriv);
>  
> +struct drm_master *drm_master_create(struct drm_device *dev);
> +
>  #endif
> diff --git a/include/drm/drm_lease.h b/include/drm/drm_lease.h
> new file mode 100644
> index ..e02adf3e42fd
> --- /dev/null
> +++ b/include/drm/drm_lease.h
> @@ -0,0 +1,51 @@
> +/*
> + * Copyright © 2017 Keith Packard 
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License as published by
> + * the Free Software Foundation, either version 2 of the License, or
> + * (at your option) any later version.
> + *
> + * This program is distributed in the hope that it will be useful, but
> + * WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> + * General Public License for more details.
> + */
> +
> +#ifndef _DRM_LEASE_H_
> +#define _DRM_LEASE_H_
> +
> +struct drm_file;
> +struct drm_device;
> +
> +struct drm_master *drm_lease_owner(struct drm_master *master);
> +
> +struct drm_master *drm_lessee_find(struct drm_master *top, int lessee_id);
> +
> +void drm_lease_destroy(struct drm_master *lessee);
> +
> +struct drm_mode_object *drm_lease_find(struct drm_master *master, int id);
> +
> +/**
> + * drm_lease_check - check drm_mode_object lease status
> + * @master: the drm_master
> + * @id: the object id
> + *
> + * Checks if the specified master holds a lease on the object
> + * and also whether it has been leased to some lessee of the
> + * specified

[PATCH 2/4] drm: Add drm_object lease infrastructure

2017-04-01 Thread Keith Packard 
This provides new data structures to hold "lease" information about
drm mode setting objects, and provides for creating new drm_masters
which have access to a subset of the available drm resources.

An 'owner' is a drm_master which is not leasing the objects from
another drm_master, and hence 'owns' them. This sits at the top of a
tree of drm_masters.

A 'lessee' is a drm_master which is leasing objects from some other
drm_master. Each lessee holds the set of objects which it is leasing
from the lessor.

A 'lessor' is a drm_master which is leasing objects to another
drm_master.

The set of objects any drm_master 'controls' is limited to the set of
objects it leases (for lessees) or all objects (for owners),
optionally minus the set of objects it has leased to other
drm_masters.

Objects not controlled by a drm_master cannot be modified through the
various state manipulating ioctls, and any state reported back to user
space will be edited to make them appear idle and/or unusable. For
instance, connectors always report 'disconnected', while encoders
report no possible crtcs or clones.

The full list of lessees leasing objects from an owner (either
directly, or indirectly through another lessee), can be searched from
an idr in the drm_master of the owner.

Signed-off-by: Keith Packard 
---
 drivers/gpu/drm/Makefile|   3 +-
 drivers/gpu/drm/drm_auth.c  |  22 +-
 drivers/gpu/drm/drm_lease.c | 485 
 include/drm/drmP.h  |   1 +
 include/drm/drm_auth.h  |  28 +++
 include/drm/drm_lease.h |  51 +
 6 files changed, 588 insertions(+), 2 deletions(-)
 create mode 100644 drivers/gpu/drm/drm_lease.c
 create mode 100644 include/drm/drm_lease.h

diff --git a/drivers/gpu/drm/Makefile b/drivers/gpu/drm/Makefile
index b9ae4280de9d..c2c6d61d30cf 100644
--- a/drivers/gpu/drm/Makefile
+++ b/drivers/gpu/drm/Makefile
@@ -16,7 +16,8 @@ drm-y   :=drm_auth.o drm_bufs.o drm_cache.o \
drm_framebuffer.o drm_connector.o drm_blend.o \
drm_encoder.o drm_mode_object.o drm_property.o \
drm_plane.o drm_color_mgmt.o drm_print.o \
-   drm_dumb_buffers.o drm_mode_config.o
+   drm_dumb_buffers.o drm_mode_config.o \
+   drm_lease.o
 
 drm-$(CONFIG_COMPAT) += drm_ioc32.o
 drm-$(CONFIG_DRM_GEM_CMA_HELPER) += drm_gem_cma_helper.o
diff --git a/drivers/gpu/drm/drm_auth.c b/drivers/gpu/drm/drm_auth.c
index 6b143514a566..1db4f63860d1 100644
--- a/drivers/gpu/drm/drm_auth.c
+++ b/drivers/gpu/drm/drm_auth.c
@@ -31,6 +31,7 @@
 #include 
 #include "drm_internal.h"
 #include "drm_legacy.h"
+#include 
 
 /**
  * DOC: master and authentication
@@ -93,7 +94,7 @@ int drm_authmagic(struct drm_device *dev, void *data,
return file ? 0 : -EINVAL;
 }
 
-static struct drm_master *drm_master_create(struct drm_device *dev)
+struct drm_master *drm_master_create(struct drm_device *dev)
 {
struct drm_master *master;
 
@@ -107,6 +108,14 @@ static struct drm_master *drm_master_create(struct 
drm_device *dev)
idr_init(>magic_map);
master->dev = dev;
 
+   /* initialize the tree of output resource lessees */
+   master->lessor = NULL;
+   master->lessee_id = 0;
+   INIT_LIST_HEAD(>lessees);
+   INIT_LIST_HEAD(>lessee_list);
+   idr_init(>leases);
+   idr_init(>lessee_idr);
+
return master;
 }
 
@@ -189,6 +198,12 @@ int drm_setmaster_ioctl(struct drm_device *dev, void *data,
goto out_unlock;
}
 
+   if (file_priv->master->lessor != NULL) {
+   DRM_DEBUG_LEASE("Attempt to set lessee %d as master\n", 
file_priv->master->lessee_id);
+   ret = -EINVAL;
+   goto out_unlock;
+   }
+
ret = drm_set_master(dev, file_priv, false);
 out_unlock:
mutex_unlock(>master_mutex);
@@ -310,12 +325,17 @@ static void drm_master_destroy(struct kref *kref)
struct drm_master *master = container_of(kref, struct drm_master, 
refcount);
struct drm_device *dev = master->dev;
 
+   drm_lease_destroy(master);
+
if (dev->driver->master_destroy)
dev->driver->master_destroy(dev, master);
 
drm_legacy_master_rmmaps(dev, master);
 
idr_destroy(>magic_map);
+
+   idr_destroy(>lessee_idr);
+
kfree(master->unique);
kfree(master);
 }
diff --git a/drivers/gpu/drm/drm_lease.c b/drivers/gpu/drm/drm_lease.c
new file mode 100644
index ..782005c7706d
--- /dev/null
+++ b/drivers/gpu/drm/drm_lease.c
@@ -0,0 +1,485 @@
+/*
+ * Copyright © 2017 Keith Packard 
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful, but

[PATCH 2/4] drm: Add drm_object lease infrastructure

2017-04-01 Thread Keith Packard 
This provides new data structures to hold "lease" information about
drm mode setting objects, and provides for creating new drm_masters
which have access to a subset of the available drm resources.

An 'owner' is a drm_master which is not leasing the objects from
another drm_master, and hence 'owns' them. This sits at the top of a
tree of drm_masters.

A 'lessee' is a drm_master which is leasing objects from some other
drm_master. Each lessee holds the set of objects which it is leasing
from the lessor.

A 'lessor' is a drm_master which is leasing objects to another
drm_master.

The set of objects any drm_master 'controls' is limited to the set of
objects it leases (for lessees) or all objects (for owners),
optionally minus the set of objects it has leased to other
drm_masters.

Objects not controlled by a drm_master cannot be modified through the
various state manipulating ioctls, and any state reported back to user
space will be edited to make them appear idle and/or unusable. For
instance, connectors always report 'disconnected', while encoders
report no possible crtcs or clones.

The full list of lessees leasing objects from an owner (either
directly, or indirectly through another lessee), can be searched from
an idr in the drm_master of the owner.

Signed-off-by: Keith Packard 
---
 drivers/gpu/drm/Makefile|   3 +-
 drivers/gpu/drm/drm_auth.c  |  22 +-
 drivers/gpu/drm/drm_lease.c | 485 
 include/drm/drmP.h  |   1 +
 include/drm/drm_auth.h  |  28 +++
 include/drm/drm_lease.h |  51 +
 6 files changed, 588 insertions(+), 2 deletions(-)
 create mode 100644 drivers/gpu/drm/drm_lease.c
 create mode 100644 include/drm/drm_lease.h

diff --git a/drivers/gpu/drm/Makefile b/drivers/gpu/drm/Makefile
index b9ae4280de9d..c2c6d61d30cf 100644
--- a/drivers/gpu/drm/Makefile
+++ b/drivers/gpu/drm/Makefile
@@ -16,7 +16,8 @@ drm-y   :=drm_auth.o drm_bufs.o drm_cache.o \
drm_framebuffer.o drm_connector.o drm_blend.o \
drm_encoder.o drm_mode_object.o drm_property.o \
drm_plane.o drm_color_mgmt.o drm_print.o \
-   drm_dumb_buffers.o drm_mode_config.o
+   drm_dumb_buffers.o drm_mode_config.o \
+   drm_lease.o
 
 drm-$(CONFIG_COMPAT) += drm_ioc32.o
 drm-$(CONFIG_DRM_GEM_CMA_HELPER) += drm_gem_cma_helper.o
diff --git a/drivers/gpu/drm/drm_auth.c b/drivers/gpu/drm/drm_auth.c
index 6b143514a566..1db4f63860d1 100644
--- a/drivers/gpu/drm/drm_auth.c
+++ b/drivers/gpu/drm/drm_auth.c
@@ -31,6 +31,7 @@
 #include 
 #include "drm_internal.h"
 #include "drm_legacy.h"
+#include 
 
 /**
  * DOC: master and authentication
@@ -93,7 +94,7 @@ int drm_authmagic(struct drm_device *dev, void *data,
return file ? 0 : -EINVAL;
 }
 
-static struct drm_master *drm_master_create(struct drm_device *dev)
+struct drm_master *drm_master_create(struct drm_device *dev)
 {
struct drm_master *master;
 
@@ -107,6 +108,14 @@ static struct drm_master *drm_master_create(struct 
drm_device *dev)
idr_init(>magic_map);
master->dev = dev;
 
+   /* initialize the tree of output resource lessees */
+   master->lessor = NULL;
+   master->lessee_id = 0;
+   INIT_LIST_HEAD(>lessees);
+   INIT_LIST_HEAD(>lessee_list);
+   idr_init(>leases);
+   idr_init(>lessee_idr);
+
return master;
 }
 
@@ -189,6 +198,12 @@ int drm_setmaster_ioctl(struct drm_device *dev, void *data,
goto out_unlock;
}
 
+   if (file_priv->master->lessor != NULL) {
+   DRM_DEBUG_LEASE("Attempt to set lessee %d as master\n", 
file_priv->master->lessee_id);
+   ret = -EINVAL;
+   goto out_unlock;
+   }
+
ret = drm_set_master(dev, file_priv, false);
 out_unlock:
mutex_unlock(>master_mutex);
@@ -310,12 +325,17 @@ static void drm_master_destroy(struct kref *kref)
struct drm_master *master = container_of(kref, struct drm_master, 
refcount);
struct drm_device *dev = master->dev;
 
+   drm_lease_destroy(master);
+
if (dev->driver->master_destroy)
dev->driver->master_destroy(dev, master);
 
drm_legacy_master_rmmaps(dev, master);
 
idr_destroy(>magic_map);
+
+   idr_destroy(>lessee_idr);
+
kfree(master->unique);
kfree(master);
 }
diff --git a/drivers/gpu/drm/drm_lease.c b/drivers/gpu/drm/drm_lease.c
new file mode 100644
index ..782005c7706d
--- /dev/null
+++ b/drivers/gpu/drm/drm_lease.c
@@ -0,0 +1,485 @@
+/*
+ * Copyright © 2017 Keith Packard 
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even