Userland interfaces

The DRM core exports several interfaces to applications, generally intended to be used through corresponding libdrm wrapper functions. In addition, drivers export device-specific interfaces for use by userspace drivers & device-aware applications through ioctls and sysfs files.

External interfaces include: memory mapping, context management, DMA operations, AGP management, vblank control, fence management, memory management, and output management.

Cover generic ioctls and sysfs layout here. We only need high-level info, since man pages should cover the rest.

libdrm Device Lookup


In an attempt to warn anyone else who’s trying to figure out what’s going on here, I’ll try to summarize the story. First things first, let’s clear up the names, because the kernel internals, libdrm and the ioctls are all named differently:

  • GET_UNIQUE ioctl, implemented by drm_getunique is wrapped up in libdrm through the drmGetBusid function.
  • The libdrm drmSetBusid function is backed by the SET_UNIQUE ioctl. All that code is nerved in the kernel with drm_invalid_op().
  • The internal set_busid kernel functions and driver callbacks are exclusively use by the SET_VERSION ioctl, because only drm 1.0 (which is nerved) allowed userspace to set the busid through the above ioctl.
  • Other ioctls and functions involved are named consistently.

For anyone wondering what’s the difference between drm 1.1 and 1.4: Correctly handling pci domains in the busid on ppc. Doing this correctly was only implemented in libdrm in 2010, hence can’t be nerved yet. No one knows what’s special with drm 1.2 and 1.3.

Now the actual horror story of how device lookup in drm works. At large, there’s 2 different ways, either by busid, or by device driver name.

Opening by busid is fairly simple:

  1. First call SET_VERSION to make sure pci domains are handled properly. As a side-effect this fills out the unique name in the master structure.
  2. Call GET_UNIQUE to read out the unique name from the master structure, which matches the busid thanks to step 1. If it doesn’t, proceed to try the next device node.

Opening by name is slightly different:

  1. Directly call VERSION to get the version and to match against the driver name returned by that ioctl. Note that SET_VERSION is not called, which means the the unique name for the master node just opening is _not_ filled out. This despite that with current drm device nodes are always bound to one device, and can’t be runtime assigned like with drm 1.0.
  2. Match driver name. If it mismatches, proceed to the next device node.
  3. Call GET_UNIQUE, and check whether the unique name has length zero (by checking that the first byte in the string is 0). If that’s not the case libdrm skips and proceeds to the next device node. Probably this is just copypasta from drm 1.0 times where a set unique name meant that the driver was in use already, but that’s just conjecture.

Long story short: To keep the open by name logic working, GET_UNIQUE must _not_ return a unique string when SET_VERSION hasn’t been called yet, otherwise libdrm breaks. Even when that unique string can’t ever change, and is totally irrelevant for actually opening the device because runtime assignable device instances were only support in drm 1.0, which is long dead. But the libdrm code in drmOpenByName somehow survived, hence this can’t be broken.

Primary Nodes, DRM Master and Authentication

struct drm_master is used to track groups of clients with open primary/legacy device nodes. For every struct drm_file which has had at least once successfully became the device master (either through the SET_MASTER IOCTL, or implicitly through opening the primary device node when no one else is the current master that time) there exists one drm_master. This is noted in the is_master member of drm_file. All other clients have just a pointer to the drm_master they are associated with.

In addition only one drm_master can be the current master for a drm_device. It can be switched through the DROP_MASTER and SET_MASTER IOCTL, or implicitly through closing/openeing the primary device node. See also drm_is_current_master().

Clients can authenticate against the current master (if it matches their own) using the GETMAGIC and AUTHMAGIC IOCTLs. Together with exchanging masters, this allows controlled access to the device for an entire group of mutually trusted clients.

bool drm_is_current_master(struct drm_file * fpriv)

checks whether priv is the current master


struct drm_file * fpriv
DRM file private


Checks whether fpriv is current master on its device. This decides whether a client is allowed to run DRM_MASTER IOCTLs.

Most of the modern IOCTL which require DRM_MASTER are for kernel modesetting - the current master is assumed to own the non-shareable display hardware.

struct drm_master * drm_master_get(struct drm_master * master)

reference a master pointer


struct drm_master * master
struct drm_master


Increments the reference count of master and returns a pointer to master.

void drm_master_put(struct drm_master ** master)

unreference and clear a master pointer


struct drm_master ** master
pointer to a pointer of struct drm_master


This decrements the drm_master behind master and sets it to NULL.

struct drm_master

drm master structure


struct drm_master {
  struct kref refcount;
  struct drm_device * dev;
  char * unique;
  int unique_len;
  struct idr magic_map;
  struct drm_lock_data lock;
  void * driver_priv;


Refcount for this master object.
Link back to the DRM device
Unique identifier: e.g. busid. Protected by drm_global_mutex.
Length of unique field. Protected by drm_global_mutex.
Map of used authentication tokens. Protected by struct_mutex.
DRI lock information.
Pointer to driver-private information.


Note that master structures are only relevant for the legacy/primary device nodes, hence there can only be one per device, not one per drm_minor.

Open-Source Userspace Requirements

The DRM subsystem has stricter requirements than most other kernel subsystems on what the userspace side for new uAPI needs to look like. This section here explains what exactly those requirements are, and why they exist.

The short summary is that any addition of DRM uAPI requires corresponding open-sourced userspace patches, and those patches must be reviewed and ready for merging into a suitable and canonical upstream project.

GFX devices (both display and render/GPU side) are really complex bits of hardware, with userspace and kernel by necessity having to work together really closely. The interfaces, for rendering and modesetting, must be extremely wide and flexible, and therefore it is almost always impossible to precisely define them for every possible corner case. This in turn makes it really practically infeasible to differentiate between behaviour that’s required by userspace, and which must not be changed to avoid regressions, and behaviour which is only an accidental artifact of the current implementation.

Without access to the full source code of all userspace users that means it becomes impossible to change the implementation details, since userspace could depend upon the accidental behaviour of the current implementation in minute details. And debugging such regressions without access to source code is pretty much impossible. As a consequence this means:

  • The Linux kernel’s “no regression” policy holds in practice only for open-source userspace of the DRM subsystem. DRM developers are perfectly fine if closed-source blob drivers in userspace use the same uAPI as the open drivers, but they must do so in the exact same way as the open drivers. Creative (ab)use of the interfaces will, and in the past routinely has, lead to breakage.
  • Any new userspace interface must have an open-source implementation as demonstration vehicle.

The other reason for requiring open-source userspace is uAPI review. Since the kernel and userspace parts of a GFX stack must work together so closely, code review can only assess whether a new interface achieves its goals by looking at both sides. Making sure that the interface indeed covers the use-case fully leads to a few additional requirements:

  • The open-source userspace must not be a toy/test application, but the real thing. Specifically it needs to handle all the usual error and corner cases. These are often the places where new uAPI falls apart and hence essential to assess the fitness of a proposed interface.
  • The userspace side must be fully reviewed and tested to the standards of that userspace project. For e.g. mesa this means piglit testcases and review on the mailing list. This is again to ensure that the new interface actually gets the job done.
  • The userspace patches must be against the canonical upstream, not some vendor fork. This is to make sure that no one cheats on the review and testing requirements by doing a quick fork.
  • The kernel patch can only be merged after all the above requirements are met, but it must be merged before the userspace patches land. uAPI always flows from the kernel, doing things the other way round risks divergence of the uAPI definitions and header files.

These are fairly steep requirements, but have grown out from years of shared pain and experience with uAPI added hastily, and almost always regretted about just as fast. GFX devices change really fast, requiring a paradigm shift and entire new set of uAPI interfaces every few years at least. Together with the Linux kernel’s guarantee to keep existing userspace running for 10+ years this is already rather painful for the DRM subsystem, with multiple different uAPIs for the same thing co-existing. If we add a few more complete mistakes into the mix every year it would be entirely unmanageable.

Render nodes

DRM core provides multiple character-devices for user-space to use. Depending on which device is opened, user-space can perform a different set of operations (mainly ioctls). The primary node is always created and called card<num>. Additionally, a currently unused control node, called controlD<num> is also created. The primary node provides all legacy operations and historically was the only interface used by userspace. With KMS, the control node was introduced. However, the planned KMS control interface has never been written and so the control node stays unused to date.

With the increased use of offscreen renderers and GPGPU applications, clients no longer require running compositors or graphics servers to make use of a GPU. But the DRM API required unprivileged clients to authenticate to a DRM-Master prior to getting GPU access. To avoid this step and to grant clients GPU access without authenticating, render nodes were introduced. Render nodes solely serve render clients, that is, no modesetting or privileged ioctls can be issued on render nodes. Only non-global rendering commands are allowed. If a driver supports render nodes, it must advertise it via the DRIVER_RENDER DRM driver capability. If not supported, the primary node must be used for render clients together with the legacy drmAuth authentication procedure.

If a driver advertises render node support, DRM core will create a separate render node called renderD<num>. There will be one render node per device. No ioctls except PRIME-related ioctls will be allowed on this node. Especially GEM_OPEN will be explicitly prohibited. Render nodes are designed to avoid the buffer-leaks, which occur if clients guess the flink names or mmap offsets on the legacy interface. Additionally to this basic interface, drivers must mark their driver-dependent render-only ioctls as DRM_RENDER_ALLOW so render clients can use them. Driver authors must be careful not to allow any privileged ioctls on render nodes.

With render nodes, user-space can now control access to the render node via basic file-system access-modes. A running graphics server which authenticates clients on the privileged primary/legacy node is no longer required. Instead, a client can open the render node and is immediately granted GPU access. Communication between clients (or servers) is done via PRIME. FLINK from render node to legacy node is not supported. New clients must not use the insecure FLINK interface.

Besides dropping all modeset/global ioctls, render nodes also drop the DRM-Master concept. There is no reason to associate render clients with a DRM-Master as they are independent of any graphics server. Besides, they must work without any running master, anyway. Drivers must be able to run without a master object if they support render nodes. If, on the other hand, a driver requires shared state between clients which is visible to user-space and accessible beyond open-file boundaries, they cannot support render nodes.

Validating changes with IGT

There’s a collection of tests that aims to cover the whole functionality of DRM drivers and that can be used to check that changes to DRM drivers or the core don’t regress existing functionality. This test suite is called IGT and its code can be found in

To build IGT, start by installing its build dependencies. In Debian-based systems:

# apt-get build-dep intel-gpu-tools

And in Fedora-based systems:

# dnf builddep intel-gpu-tools

Then clone the repository:

$ git clone git://

Configure the build system and start the build:

$ cd igt-gpu-tools && ./ && make -j6

Download the piglit dependency:

$ ./scripts/ -d

And run the tests:

$ ./scripts/ -t kms -t core -s is a wrapper around piglit that will execute the tests matching the -t options. A report in HTML format will be available in ./results/html/index.html. Results can be compared with piglit.

VBlank event handling

The DRM core exposes two vertical blank related ioctls:

This takes a struct drm_wait_vblank structure as its argument, and it is used to block or request a signal when a specified vblank event occurs.
This was only used for user-mode-settind drivers around modesetting changes to allow the kernel to update the vblank interrupt after mode setting, since on many devices the vertical blank counter is reset to 0 at some point during modeset. Modern drivers should not call this any more since with kernel mode setting it is a no-op.

This second part of the GPU Driver Developer’s Guide documents driver code, implementation details and also all the driver-specific userspace interfaces. Especially since all hardware-acceleration interfaces to userspace are driver specific for efficiency and other reasons these interfaces can be rather substantial. Hence every driver has its own chapter.