1 ===================== 1 =====================
2 DRM Memory Management 2 DRM Memory Management
3 ===================== 3 =====================
4 4
5 Modern Linux systems require large amount of g 5 Modern Linux systems require large amount of graphics memory to store
6 frame buffers, textures, vertices and other gr 6 frame buffers, textures, vertices and other graphics-related data. Given
7 the very dynamic nature of many of that data, 7 the very dynamic nature of many of that data, managing graphics memory
8 efficiently is thus crucial for the graphics s 8 efficiently is thus crucial for the graphics stack and plays a central
9 role in the DRM infrastructure. 9 role in the DRM infrastructure.
10 10
11 The DRM core includes two memory managers, nam !! 11 The DRM core includes two memory managers, namely Translation Table Maps
12 (TTM) and Graphics Execution Manager (GEM). TT 12 (TTM) and Graphics Execution Manager (GEM). TTM was the first DRM memory
13 manager to be developed and tried to be a one- 13 manager to be developed and tried to be a one-size-fits-them all
14 solution. It provides a single userspace API t 14 solution. It provides a single userspace API to accommodate the need of
15 all hardware, supporting both Unified Memory A 15 all hardware, supporting both Unified Memory Architecture (UMA) devices
16 and devices with dedicated video RAM (i.e. mos 16 and devices with dedicated video RAM (i.e. most discrete video cards).
17 This resulted in a large, complex piece of cod 17 This resulted in a large, complex piece of code that turned out to be
18 hard to use for driver development. 18 hard to use for driver development.
19 19
20 GEM started as an Intel-sponsored project in r 20 GEM started as an Intel-sponsored project in reaction to TTM's
21 complexity. Its design philosophy is completel 21 complexity. Its design philosophy is completely different: instead of
22 providing a solution to every graphics memory- 22 providing a solution to every graphics memory-related problems, GEM
23 identified common code between drivers and cre 23 identified common code between drivers and created a support library to
24 share it. GEM has simpler initialization and e 24 share it. GEM has simpler initialization and execution requirements than
25 TTM, but has no video RAM management capabilit 25 TTM, but has no video RAM management capabilities and is thus limited to
26 UMA devices. 26 UMA devices.
27 27
28 The Translation Table Manager (TTM) 28 The Translation Table Manager (TTM)
29 =================================== 29 ===================================
30 30
31 .. kernel-doc:: drivers/gpu/drm/ttm/ttm_module !! 31 TTM design background and information belongs here.
32 :doc: TTM <<
33 32
34 .. kernel-doc:: include/drm/ttm/ttm_caching.h !! 33 TTM initialization
35 :internal: !! 34 ------------------
36 <<
37 TTM device object reference <<
38 --------------------------- <<
39 <<
40 .. kernel-doc:: include/drm/ttm/ttm_device.h <<
41 :internal: <<
42 <<
43 .. kernel-doc:: drivers/gpu/drm/ttm/ttm_device <<
44 :export: <<
45 <<
46 TTM resource placement reference <<
47 -------------------------------- <<
48 <<
49 .. kernel-doc:: include/drm/ttm/ttm_placement. <<
50 :internal: <<
51 <<
52 TTM resource object reference <<
53 ----------------------------- <<
54 <<
55 .. kernel-doc:: include/drm/ttm/ttm_resource.h <<
56 :internal: <<
57 <<
58 .. kernel-doc:: drivers/gpu/drm/ttm/ttm_resour <<
59 :export: <<
60 <<
61 TTM TT object reference <<
62 ----------------------- <<
63 <<
64 .. kernel-doc:: include/drm/ttm/ttm_tt.h <<
65 :internal: <<
66 35
67 .. kernel-doc:: drivers/gpu/drm/ttm/ttm_tt.c !! 36 **Warning**
68 :export: <<
69 37
70 TTM page pool reference !! 38 This section is outdated.
71 ----------------------- <<
72 39
73 .. kernel-doc:: include/drm/ttm/ttm_pool.h !! 40 Drivers wishing to support TTM must fill out a drm_bo_driver
74 :internal: !! 41 structure. The structure contains several fields with function pointers
>> 42 for initializing the TTM, allocating and freeing memory, waiting for
>> 43 command completion and fence synchronization, and memory migration. See
>> 44 the radeon_ttm.c file for an example of usage.
>> 45
>> 46 The ttm_global_reference structure is made up of several fields:
>> 47
>> 48 ::
>> 49
>> 50 struct ttm_global_reference {
>> 51 enum ttm_global_types global_type;
>> 52 size_t size;
>> 53 void *object;
>> 54 int (*init) (struct ttm_global_reference *);
>> 55 void (*release) (struct ttm_global_reference *);
>> 56 };
>> 57
>> 58
>> 59 There should be one global reference structure for your memory manager
>> 60 as a whole, and there will be others for each object created by the
>> 61 memory manager at runtime. Your global TTM should have a type of
>> 62 TTM_GLOBAL_TTM_MEM. The size field for the global object should be
>> 63 sizeof(struct ttm_mem_global), and the init and release hooks should
>> 64 point at your driver-specific init and release routines, which probably
>> 65 eventually call ttm_mem_global_init and ttm_mem_global_release,
>> 66 respectively.
75 67
76 .. kernel-doc:: drivers/gpu/drm/ttm/ttm_pool.c !! 68 Once your global TTM accounting structure is set up and initialized by
77 :export: !! 69 calling ttm_global_item_ref() on it, you need to create a buffer
>> 70 object TTM to provide a pool for buffer object allocation by clients and
>> 71 the kernel itself. The type of this object should be
>> 72 TTM_GLOBAL_TTM_BO, and its size should be sizeof(struct
>> 73 ttm_bo_global). Again, driver-specific init and release functions may
>> 74 be provided, likely eventually calling ttm_bo_global_init() and
>> 75 ttm_bo_global_release(), respectively. Also, like the previous
>> 76 object, ttm_global_item_ref() is used to create an initial reference
>> 77 count for the TTM, which will call your initialization function.
78 78
79 The Graphics Execution Manager (GEM) 79 The Graphics Execution Manager (GEM)
80 ==================================== 80 ====================================
81 81
82 The GEM design approach has resulted in a memo 82 The GEM design approach has resulted in a memory manager that doesn't
83 provide full coverage of all (or even all comm 83 provide full coverage of all (or even all common) use cases in its
84 userspace or kernel API. GEM exposes a set of 84 userspace or kernel API. GEM exposes a set of standard memory-related
85 operations to userspace and a set of helper fu 85 operations to userspace and a set of helper functions to drivers, and
86 let drivers implement hardware-specific operat 86 let drivers implement hardware-specific operations with their own
87 private API. 87 private API.
88 88
89 The GEM userspace API is described in the `GEM 89 The GEM userspace API is described in the `GEM - the Graphics Execution
90 Manager <http://lwn.net/Articles/283798/>`__ a 90 Manager <http://lwn.net/Articles/283798/>`__ article on LWN. While
91 slightly outdated, the document provides a goo 91 slightly outdated, the document provides a good overview of the GEM API
92 principles. Buffer allocation and read and wri 92 principles. Buffer allocation and read and write operations, described
93 as part of the common GEM API, are currently i 93 as part of the common GEM API, are currently implemented using
94 driver-specific ioctls. 94 driver-specific ioctls.
95 95
96 GEM is data-agnostic. It manages abstract buff 96 GEM is data-agnostic. It manages abstract buffer objects without knowing
97 what individual buffers contain. APIs that req 97 what individual buffers contain. APIs that require knowledge of buffer
98 contents or purpose, such as buffer allocation 98 contents or purpose, such as buffer allocation or synchronization
99 primitives, are thus outside of the scope of G 99 primitives, are thus outside of the scope of GEM and must be implemented
100 using driver-specific ioctls. 100 using driver-specific ioctls.
101 101
102 On a fundamental level, GEM involves several o 102 On a fundamental level, GEM involves several operations:
103 103
104 - Memory allocation and freeing 104 - Memory allocation and freeing
105 - Command execution 105 - Command execution
106 - Aperture management at command execution ti 106 - Aperture management at command execution time
107 107
108 Buffer object allocation is relatively straigh 108 Buffer object allocation is relatively straightforward and largely
109 provided by Linux's shmem layer, which provide 109 provided by Linux's shmem layer, which provides memory to back each
110 object. 110 object.
111 111
112 Device-specific operations, such as command ex 112 Device-specific operations, such as command execution, pinning, buffer
113 read & write, mapping, and domain ownership tr 113 read & write, mapping, and domain ownership transfers are left to
114 driver-specific ioctls. 114 driver-specific ioctls.
115 115
116 GEM Initialization 116 GEM Initialization
117 ------------------ 117 ------------------
118 118
119 Drivers that use GEM must set the DRIVER_GEM b 119 Drivers that use GEM must set the DRIVER_GEM bit in the struct
120 :c:type:`struct drm_driver <drm_driver>` drive 120 :c:type:`struct drm_driver <drm_driver>` driver_features
121 field. The DRM core will then automatically in 121 field. The DRM core will then automatically initialize the GEM core
122 before calling the load operation. Behind the 122 before calling the load operation. Behind the scene, this will create a
123 DRM Memory Manager object which provides an ad 123 DRM Memory Manager object which provides an address space pool for
124 object allocation. 124 object allocation.
125 125
126 In a KMS configuration, drivers need to alloca 126 In a KMS configuration, drivers need to allocate and initialize a
127 command ring buffer following core GEM initial 127 command ring buffer following core GEM initialization if required by the
128 hardware. UMA devices usually have what is cal 128 hardware. UMA devices usually have what is called a "stolen" memory
129 region, which provides space for the initial f 129 region, which provides space for the initial framebuffer and large,
130 contiguous memory regions required by the devi 130 contiguous memory regions required by the device. This space is
131 typically not managed by GEM, and must be init 131 typically not managed by GEM, and must be initialized separately into
132 its own DRM MM object. 132 its own DRM MM object.
133 133
134 GEM Objects Creation 134 GEM Objects Creation
135 -------------------- 135 --------------------
136 136
137 GEM splits creation of GEM objects and allocat 137 GEM splits creation of GEM objects and allocation of the memory that
138 backs them in two distinct operations. 138 backs them in two distinct operations.
139 139
140 GEM objects are represented by an instance of 140 GEM objects are represented by an instance of struct :c:type:`struct
141 drm_gem_object <drm_gem_object>`. Drivers usua 141 drm_gem_object <drm_gem_object>`. Drivers usually need to
142 extend GEM objects with private information an 142 extend GEM objects with private information and thus create a
143 driver-specific GEM object structure type that 143 driver-specific GEM object structure type that embeds an instance of
144 struct :c:type:`struct drm_gem_object <drm_gem 144 struct :c:type:`struct drm_gem_object <drm_gem_object>`.
145 145
146 To create a GEM object, a driver allocates mem 146 To create a GEM object, a driver allocates memory for an instance of its
147 specific GEM object type and initializes the e 147 specific GEM object type and initializes the embedded struct
148 :c:type:`struct drm_gem_object <drm_gem_object 148 :c:type:`struct drm_gem_object <drm_gem_object>` with a call
149 to drm_gem_object_init(). The function takes a !! 149 to :c:func:`drm_gem_object_init()`. The function takes a pointer
150 to the DRM device, a pointer to the GEM object 150 to the DRM device, a pointer to the GEM object and the buffer object
151 size in bytes. 151 size in bytes.
152 152
153 GEM uses shmem to allocate anonymous pageable 153 GEM uses shmem to allocate anonymous pageable memory.
154 drm_gem_object_init() will create an shmfs fil !! 154 :c:func:`drm_gem_object_init()` will create an shmfs file of the
155 requested size and store it into the struct :c 155 requested size and store it into the struct :c:type:`struct
156 drm_gem_object <drm_gem_object>` filp field. T 156 drm_gem_object <drm_gem_object>` filp field. The memory is
157 used as either main storage for the object whe 157 used as either main storage for the object when the graphics hardware
158 uses system memory directly or as a backing st 158 uses system memory directly or as a backing store otherwise.
159 159
160 Drivers are responsible for the actual physica 160 Drivers are responsible for the actual physical pages allocation by
161 calling shmem_read_mapping_page_gfp() for each !! 161 calling :c:func:`shmem_read_mapping_page_gfp()` for each page.
162 Note that they can decide to allocate pages wh 162 Note that they can decide to allocate pages when initializing the GEM
163 object, or to delay allocation until the memor 163 object, or to delay allocation until the memory is needed (for instance
164 when a page fault occurs as a result of a user 164 when a page fault occurs as a result of a userspace memory access or
165 when the driver needs to start a DMA transfer 165 when the driver needs to start a DMA transfer involving the memory).
166 166
167 Anonymous pageable memory allocation is not al 167 Anonymous pageable memory allocation is not always desired, for instance
168 when the hardware requires physically contiguo 168 when the hardware requires physically contiguous system memory as is
169 often the case in embedded devices. Drivers ca 169 often the case in embedded devices. Drivers can create GEM objects with
170 no shmfs backing (called private GEM objects) !! 170 no shmfs backing (called private GEM objects) by initializing them with
171 to drm_gem_private_object_init() instead of dr !! 171 a call to :c:func:`drm_gem_private_object_init()` instead of
172 private GEM objects must be managed by drivers !! 172 :c:func:`drm_gem_object_init()`. Storage for private GEM objects
>> 173 must be managed by drivers.
173 174
174 GEM Objects Lifetime 175 GEM Objects Lifetime
175 -------------------- 176 --------------------
176 177
177 All GEM objects are reference-counted by the G 178 All GEM objects are reference-counted by the GEM core. References can be
178 acquired and release by calling drm_gem_object !! 179 acquired and release by :c:func:`calling
179 respectively. !! 180 drm_gem_object_reference()` and
>> 181 :c:func:`drm_gem_object_unreference()` respectively. The caller
>> 182 must hold the :c:type:`struct drm_device <drm_device>`
>> 183 struct_mutex lock when calling
>> 184 :c:func:`drm_gem_object_reference()`. As a convenience, GEM
>> 185 provides :c:func:`drm_gem_object_unreference_unlocked()`
>> 186 functions that can be called without holding the lock.
180 187
181 When the last reference to a GEM object is rel 188 When the last reference to a GEM object is released the GEM core calls
182 the :c:type:`struct drm_gem_object_funcs <gem_ !! 189 the :c:type:`struct drm_driver <drm_driver>` gem_free_object
183 operation. That operation is mandatory for GEM 190 operation. That operation is mandatory for GEM-enabled drivers and must
184 free the GEM object and all associated resourc 191 free the GEM object and all associated resources.
185 192
186 void (\*free) (struct drm_gem_object \*obj); D !! 193 void (\*gem_free_object) (struct drm_gem_object \*obj); Drivers are
187 responsible for freeing all GEM object resourc 194 responsible for freeing all GEM object resources. This includes the
188 resources created by the GEM core, which need 195 resources created by the GEM core, which need to be released with
189 drm_gem_object_release(). !! 196 :c:func:`drm_gem_object_release()`.
190 197
191 GEM Objects Naming 198 GEM Objects Naming
192 ------------------ 199 ------------------
193 200
194 Communication between userspace and the kernel 201 Communication between userspace and the kernel refers to GEM objects
195 using local handles, global names or, more rec 202 using local handles, global names or, more recently, file descriptors.
196 All of those are 32-bit integer values; the us 203 All of those are 32-bit integer values; the usual Linux kernel limits
197 apply to the file descriptors. 204 apply to the file descriptors.
198 205
199 GEM handles are local to a DRM file. Applicati 206 GEM handles are local to a DRM file. Applications get a handle to a GEM
200 object through a driver-specific ioctl, and ca 207 object through a driver-specific ioctl, and can use that handle to refer
201 to the GEM object in other standard or driver- 208 to the GEM object in other standard or driver-specific ioctls. Closing a
202 DRM file handle frees all its GEM handles and 209 DRM file handle frees all its GEM handles and dereferences the
203 associated GEM objects. 210 associated GEM objects.
204 211
205 To create a handle for a GEM object drivers ca !! 212 To create a handle for a GEM object drivers call
206 function takes a pointer to the DRM file and t !! 213 :c:func:`drm_gem_handle_create()`. The function takes a pointer
207 locally unique handle. When the handle is no !! 214 to the DRM file and the GEM object and returns a locally unique handle.
208 with a call to drm_gem_handle_delete(). Finall !! 215 When the handle is no longer needed drivers delete it with a call to
209 handle can be retrieved by a call to drm_gem_o !! 216 :c:func:`drm_gem_handle_delete()`. Finally the GEM object
>> 217 associated with a handle can be retrieved by a call to
>> 218 :c:func:`drm_gem_object_lookup()`.
210 219
211 Handles don't take ownership of GEM objects, t 220 Handles don't take ownership of GEM objects, they only take a reference
212 to the object that will be dropped when the ha 221 to the object that will be dropped when the handle is destroyed. To
213 avoid leaking GEM objects, drivers must make s 222 avoid leaking GEM objects, drivers must make sure they drop the
214 reference(s) they own (such as the initial ref 223 reference(s) they own (such as the initial reference taken at object
215 creation time) as appropriate, without any spe 224 creation time) as appropriate, without any special consideration for the
216 handle. For example, in the particular case of 225 handle. For example, in the particular case of combined GEM object and
217 handle creation in the implementation of the d 226 handle creation in the implementation of the dumb_create operation,
218 drivers must drop the initial reference to the 227 drivers must drop the initial reference to the GEM object before
219 returning the handle. 228 returning the handle.
220 229
221 GEM names are similar in purpose to handles bu 230 GEM names are similar in purpose to handles but are not local to DRM
222 files. They can be passed between processes to 231 files. They can be passed between processes to reference a GEM object
223 globally. Names can't be used directly to refe 232 globally. Names can't be used directly to refer to objects in the DRM
224 API, applications must convert handles to name 233 API, applications must convert handles to names and names to handles
225 using the DRM_IOCTL_GEM_FLINK and DRM_IOCTL_GE 234 using the DRM_IOCTL_GEM_FLINK and DRM_IOCTL_GEM_OPEN ioctls
226 respectively. The conversion is handled by the 235 respectively. The conversion is handled by the DRM core without any
227 driver-specific support. 236 driver-specific support.
228 237
229 GEM also supports buffer sharing with dma-buf 238 GEM also supports buffer sharing with dma-buf file descriptors through
230 PRIME. GEM-based drivers must use the provided 239 PRIME. GEM-based drivers must use the provided helpers functions to
231 implement the exporting and importing correctl 240 implement the exporting and importing correctly. See ?. Since sharing
232 file descriptors is inherently more secure tha 241 file descriptors is inherently more secure than the easily guessable and
233 global GEM names it is the preferred buffer sh 242 global GEM names it is the preferred buffer sharing mechanism. Sharing
234 buffers through GEM names is only supported fo 243 buffers through GEM names is only supported for legacy userspace.
235 Furthermore PRIME also allows cross-device buf 244 Furthermore PRIME also allows cross-device buffer sharing since it is
236 based on dma-bufs. 245 based on dma-bufs.
237 246
238 GEM Objects Mapping 247 GEM Objects Mapping
239 ------------------- 248 -------------------
240 249
241 Because mapping operations are fairly heavywei 250 Because mapping operations are fairly heavyweight GEM favours
242 read/write-like access to buffers, implemented 251 read/write-like access to buffers, implemented through driver-specific
243 ioctls, over mapping buffers to userspace. How 252 ioctls, over mapping buffers to userspace. However, when random access
244 to the buffer is needed (to perform software r 253 to the buffer is needed (to perform software rendering for instance),
245 direct access to the object can be more effici 254 direct access to the object can be more efficient.
246 255
247 The mmap system call can't be used directly to 256 The mmap system call can't be used directly to map GEM objects, as they
248 don't have their own file handle. Two alternat 257 don't have their own file handle. Two alternative methods currently
249 co-exist to map GEM objects to userspace. The 258 co-exist to map GEM objects to userspace. The first method uses a
250 driver-specific ioctl to perform the mapping o 259 driver-specific ioctl to perform the mapping operation, calling
251 do_mmap() under the hood. This is often consid !! 260 :c:func:`do_mmap()` under the hood. This is often considered
252 dubious, seems to be discouraged for new GEM-e 261 dubious, seems to be discouraged for new GEM-enabled drivers, and will
253 thus not be described here. 262 thus not be described here.
254 263
255 The second method uses the mmap system call on 264 The second method uses the mmap system call on the DRM file handle. void
256 \*mmap(void \*addr, size_t length, int prot, i 265 \*mmap(void \*addr, size_t length, int prot, int flags, int fd, off_t
257 offset); DRM identifies the GEM object to be m 266 offset); DRM identifies the GEM object to be mapped by a fake offset
258 passed through the mmap offset argument. Prior 267 passed through the mmap offset argument. Prior to being mapped, a GEM
259 object must thus be associated with a fake off 268 object must thus be associated with a fake offset. To do so, drivers
260 must call drm_gem_create_mmap_offset() on the !! 269 must call :c:func:`drm_gem_create_mmap_offset()` on the object.
261 270
262 Once allocated, the fake offset value must be 271 Once allocated, the fake offset value must be passed to the application
263 in a driver-specific way and can then be used 272 in a driver-specific way and can then be used as the mmap offset
264 argument. 273 argument.
265 274
266 The GEM core provides a helper method drm_gem_ !! 275 The GEM core provides a helper method :c:func:`drm_gem_mmap()` to
267 handle object mapping. The method can be set d 276 handle object mapping. The method can be set directly as the mmap file
268 operation handler. It will look up the GEM obj 277 operation handler. It will look up the GEM object based on the offset
269 value and set the VMA operations to the :c:typ 278 value and set the VMA operations to the :c:type:`struct drm_driver
270 <drm_driver>` gem_vm_ops field. Note that drm_ !! 279 <drm_driver>` gem_vm_ops field. Note that
271 userspace, but relies on the driver-provided f !! 280 :c:func:`drm_gem_mmap()` doesn't map memory to userspace, but
272 individually. !! 281 relies on the driver-provided fault handler to map pages individually.
273 !! 282
274 To use drm_gem_mmap(), drivers must fill the s !! 283 To use :c:func:`drm_gem_mmap()`, drivers must fill the struct
275 <drm_driver>` gem_vm_ops field with a pointer !! 284 :c:type:`struct drm_driver <drm_driver>` gem_vm_ops field
276 !! 285 with a pointer to VM operations.
277 The VM operations is a :c:type:`struct vm_oper !! 286
278 made up of several fields, the more interestin !! 287 struct vm_operations_struct \*gem_vm_ops struct
279 !! 288 vm_operations_struct { void (\*open)(struct vm_area_struct \* area);
280 .. code-block:: c !! 289 void (\*close)(struct vm_area_struct \* area); int (\*fault)(struct
281 !! 290 vm_area_struct \*vma, struct vm_fault \*vmf); };
282 struct vm_operations_struct { <<
283 void (*open)(struct vm_area_st <<
284 void (*close)(struct vm_area_s <<
285 vm_fault_t (*fault)(struct vm_ <<
286 }; <<
287 <<
288 291
289 The open and close operations must update the 292 The open and close operations must update the GEM object reference
290 count. Drivers can use the drm_gem_vm_open() a !! 293 count. Drivers can use the :c:func:`drm_gem_vm_open()` and
291 functions directly as open and close handlers. !! 294 :c:func:`drm_gem_vm_close()` helper functions directly as open
>> 295 and close handlers.
292 296
293 The fault operation handler is responsible for 297 The fault operation handler is responsible for mapping individual pages
294 to userspace when a page fault occurs. Dependi 298 to userspace when a page fault occurs. Depending on the memory
295 allocation scheme, drivers can allocate pages 299 allocation scheme, drivers can allocate pages at fault time, or can
296 decide to allocate memory for the GEM object a 300 decide to allocate memory for the GEM object at the time the object is
297 created. 301 created.
298 302
299 Drivers that want to map the GEM object upfron 303 Drivers that want to map the GEM object upfront instead of handling page
300 faults can implement their own mmap file opera 304 faults can implement their own mmap file operation handler.
301 305
302 For platforms without MMU the GEM core provide <<
303 drm_gem_dma_get_unmapped_area(). The mmap() ro <<
304 proposed address for the mapping. <<
305 <<
306 To use drm_gem_dma_get_unmapped_area(), driver <<
307 :c:type:`struct file_operations <file_operatio <<
308 a pointer on drm_gem_dma_get_unmapped_area(). <<
309 <<
310 More detailed information about get_unmapped_a <<
311 Documentation/admin-guide/mm/nommu-mmap.rst <<
312 <<
313 Memory Coherency 306 Memory Coherency
314 ---------------- 307 ----------------
315 308
316 When mapped to the device or used in a command 309 When mapped to the device or used in a command buffer, backing pages for
317 an object are flushed to memory and marked wri 310 an object are flushed to memory and marked write combined so as to be
318 coherent with the GPU. Likewise, if the CPU ac 311 coherent with the GPU. Likewise, if the CPU accesses an object after the
319 GPU has finished rendering to the object, then 312 GPU has finished rendering to the object, then the object must be made
320 coherent with the CPU's view of memory, usuall 313 coherent with the CPU's view of memory, usually involving GPU cache
321 flushing of various kinds. This core CPU<->GPU 314 flushing of various kinds. This core CPU<->GPU coherency management is
322 provided by a device-specific ioctl, which eva 315 provided by a device-specific ioctl, which evaluates an object's current
323 domain and performs any necessary flushing or 316 domain and performs any necessary flushing or synchronization to put the
324 object into the desired coherency domain (note 317 object into the desired coherency domain (note that the object may be
325 busy, i.e. an active render target; in that ca 318 busy, i.e. an active render target; in that case, setting the domain
326 blocks the client and waits for rendering to c 319 blocks the client and waits for rendering to complete before performing
327 any necessary flushing operations). 320 any necessary flushing operations).
328 321
329 Command Execution 322 Command Execution
330 ----------------- 323 -----------------
331 324
332 Perhaps the most important GEM function for GP 325 Perhaps the most important GEM function for GPU devices is providing a
333 command execution interface to clients. Client 326 command execution interface to clients. Client programs construct
334 command buffers containing references to previ 327 command buffers containing references to previously allocated memory
335 objects, and then submit them to GEM. At that 328 objects, and then submit them to GEM. At that point, GEM takes care to
336 bind all the objects into the GTT, execute the 329 bind all the objects into the GTT, execute the buffer, and provide
337 necessary synchronization between clients acce 330 necessary synchronization between clients accessing the same buffers.
338 This often involves evicting some objects from 331 This often involves evicting some objects from the GTT and re-binding
339 others (a fairly expensive operation), and pro 332 others (a fairly expensive operation), and providing relocation support
340 which hides fixed GTT offsets from clients. Cl 333 which hides fixed GTT offsets from clients. Clients must take care not
341 to submit command buffers that reference more 334 to submit command buffers that reference more objects than can fit in
342 the GTT; otherwise, GEM will reject them and n 335 the GTT; otherwise, GEM will reject them and no rendering will occur.
343 Similarly, if several objects in the buffer re 336 Similarly, if several objects in the buffer require fence registers to
344 be allocated for correct rendering (e.g. 2D bl 337 be allocated for correct rendering (e.g. 2D blits on pre-965 chips),
345 care must be taken not to require more fence r 338 care must be taken not to require more fence registers than are
346 available to the client. Such resource managem 339 available to the client. Such resource management should be abstracted
347 from the client in libdrm. 340 from the client in libdrm.
348 341
349 GEM Function Reference 342 GEM Function Reference
350 ---------------------- 343 ----------------------
351 344
352 .. kernel-doc:: include/drm/drm_gem.h <<
353 :internal: <<
354 <<
355 .. kernel-doc:: drivers/gpu/drm/drm_gem.c 345 .. kernel-doc:: drivers/gpu/drm/drm_gem.c
356 :export: 346 :export:
357 347
358 GEM DMA Helper Functions Reference !! 348 .. kernel-doc:: include/drm/drm_gem.h
359 ---------------------------------- <<
360 <<
361 .. kernel-doc:: drivers/gpu/drm/drm_gem_dma_he <<
362 :doc: dma helpers <<
363 <<
364 .. kernel-doc:: include/drm/drm_gem_dma_helper <<
365 :internal: 349 :internal:
366 350
367 .. kernel-doc:: drivers/gpu/drm/drm_gem_dma_he !! 351 GEM CMA Helper Functions Reference
368 :export: !! 352 ----------------------------------
369 <<
370 GEM SHMEM Helper Function Reference <<
371 ----------------------------------- <<
372 353
373 .. kernel-doc:: drivers/gpu/drm/drm_gem_shmem_ !! 354 .. kernel-doc:: drivers/gpu/drm/drm_gem_cma_helper.c
374 :doc: overview !! 355 :doc: cma helpers
375 356
376 .. kernel-doc:: include/drm/drm_gem_shmem_help !! 357 .. kernel-doc:: drivers/gpu/drm/drm_gem_cma_helper.c
377 :internal: <<
378 <<
379 .. kernel-doc:: drivers/gpu/drm/drm_gem_shmem_ <<
380 :export: 358 :export:
381 359
382 GEM VRAM Helper Functions Reference !! 360 .. kernel-doc:: include/drm/drm_gem_cma_helper.h
383 ----------------------------------- <<
384 <<
385 .. kernel-doc:: drivers/gpu/drm/drm_gem_vram_h <<
386 :doc: overview <<
387 <<
388 .. kernel-doc:: include/drm/drm_gem_vram_helpe <<
389 :internal: 361 :internal:
390 362
391 .. kernel-doc:: drivers/gpu/drm/drm_gem_vram_h <<
392 :export: <<
393 <<
394 GEM TTM Helper Functions Reference <<
395 ----------------------------------- <<
396 <<
397 .. kernel-doc:: drivers/gpu/drm/drm_gem_ttm_he <<
398 :doc: overview <<
399 <<
400 .. kernel-doc:: drivers/gpu/drm/drm_gem_ttm_he <<
401 :export: <<
402 <<
403 VMA Offset Manager 363 VMA Offset Manager
404 ================== 364 ==================
405 365
406 .. kernel-doc:: drivers/gpu/drm/drm_vma_manage 366 .. kernel-doc:: drivers/gpu/drm/drm_vma_manager.c
407 :doc: vma offset manager 367 :doc: vma offset manager
408 368
409 .. kernel-doc:: include/drm/drm_vma_manager.h <<
410 :internal: <<
411 <<
412 .. kernel-doc:: drivers/gpu/drm/drm_vma_manage 369 .. kernel-doc:: drivers/gpu/drm/drm_vma_manager.c
413 :export: 370 :export:
414 371
415 .. _prime_buffer_sharing: !! 372 .. kernel-doc:: include/drm/drm_vma_manager.h
>> 373 :internal:
416 374
417 PRIME Buffer Sharing 375 PRIME Buffer Sharing
418 ==================== 376 ====================
419 377
420 PRIME is the cross device buffer sharing frame 378 PRIME is the cross device buffer sharing framework in drm, originally
421 created for the OPTIMUS range of multi-gpu pla 379 created for the OPTIMUS range of multi-gpu platforms. To userspace PRIME
422 buffers are dma-buf based file descriptors. 380 buffers are dma-buf based file descriptors.
423 381
424 Overview and Lifetime Rules !! 382 Overview and Driver Interface
425 --------------------------- !! 383 -----------------------------
426 384
427 .. kernel-doc:: drivers/gpu/drm/drm_prime.c !! 385 Similar to GEM global names, PRIME file descriptors are also used to
428 :doc: overview and lifetime rules !! 386 share buffer objects across processes. They offer additional security:
>> 387 as file descriptors must be explicitly sent over UNIX domain sockets to
>> 388 be shared between applications, they can't be guessed like the globally
>> 389 unique GEM names.
>> 390
>> 391 Drivers that support the PRIME API must set the DRIVER_PRIME bit in the
>> 392 struct :c:type:`struct drm_driver <drm_driver>`
>> 393 driver_features field, and implement the prime_handle_to_fd and
>> 394 prime_fd_to_handle operations.
>> 395
>> 396 int (\*prime_handle_to_fd)(struct drm_device \*dev, struct drm_file
>> 397 \*file_priv, uint32_t handle, uint32_t flags, int \*prime_fd); int
>> 398 (\*prime_fd_to_handle)(struct drm_device \*dev, struct drm_file
>> 399 \*file_priv, int prime_fd, uint32_t \*handle); Those two operations
>> 400 convert a handle to a PRIME file descriptor and vice versa. Drivers must
>> 401 use the kernel dma-buf buffer sharing framework to manage the PRIME file
>> 402 descriptors. Similar to the mode setting API PRIME is agnostic to the
>> 403 underlying buffer object manager, as long as handles are 32bit unsigned
>> 404 integers.
>> 405
>> 406 While non-GEM drivers must implement the operations themselves, GEM
>> 407 drivers must use the :c:func:`drm_gem_prime_handle_to_fd()` and
>> 408 :c:func:`drm_gem_prime_fd_to_handle()` helper functions. Those
>> 409 helpers rely on the driver gem_prime_export and gem_prime_import
>> 410 operations to create a dma-buf instance from a GEM object (dma-buf
>> 411 exporter role) and to create a GEM object from a dma-buf instance
>> 412 (dma-buf importer role).
>> 413
>> 414 struct dma_buf \* (\*gem_prime_export)(struct drm_device \*dev,
>> 415 struct drm_gem_object \*obj, int flags); struct drm_gem_object \*
>> 416 (\*gem_prime_import)(struct drm_device \*dev, struct dma_buf
>> 417 \*dma_buf); These two operations are mandatory for GEM drivers that
>> 418 support PRIME.
429 419
430 PRIME Helper Functions 420 PRIME Helper Functions
431 ---------------------- 421 ----------------------
432 422
433 .. kernel-doc:: drivers/gpu/drm/drm_prime.c 423 .. kernel-doc:: drivers/gpu/drm/drm_prime.c
434 :doc: PRIME Helpers 424 :doc: PRIME Helpers
435 425
436 PRIME Function References 426 PRIME Function References
437 ------------------------- 427 -------------------------
438 428
439 .. kernel-doc:: include/drm/drm_prime.h <<
440 :internal: <<
441 <<
442 .. kernel-doc:: drivers/gpu/drm/drm_prime.c 429 .. kernel-doc:: drivers/gpu/drm/drm_prime.c
443 :export: 430 :export:
444 431
445 DRM MM Range Allocator 432 DRM MM Range Allocator
446 ====================== 433 ======================
447 434
448 Overview 435 Overview
449 -------- 436 --------
450 437
451 .. kernel-doc:: drivers/gpu/drm/drm_mm.c 438 .. kernel-doc:: drivers/gpu/drm/drm_mm.c
452 :doc: Overview 439 :doc: Overview
453 440
454 LRU Scan/Eviction Support 441 LRU Scan/Eviction Support
455 ------------------------- 442 -------------------------
456 443
457 .. kernel-doc:: drivers/gpu/drm/drm_mm.c 444 .. kernel-doc:: drivers/gpu/drm/drm_mm.c
458 :doc: lru scan roster !! 445 :doc: lru scan roaster
459 446
460 DRM MM Range Allocator Function References 447 DRM MM Range Allocator Function References
461 ------------------------------------------ 448 ------------------------------------------
462 449
463 .. kernel-doc:: include/drm/drm_mm.h <<
464 :internal: <<
465 <<
466 .. kernel-doc:: drivers/gpu/drm/drm_mm.c 450 .. kernel-doc:: drivers/gpu/drm/drm_mm.c
467 :export: 451 :export:
468 452
469 .. _drm_gpuvm: !! 453 .. kernel-doc:: include/drm/drm_mm.h
470 <<
471 DRM GPUVM <<
472 ========= <<
473 <<
474 Overview <<
475 -------- <<
476 <<
477 .. kernel-doc:: drivers/gpu/drm/drm_gpuvm.c <<
478 :doc: Overview <<
479 <<
480 Split and Merge <<
481 --------------- <<
482 <<
483 .. kernel-doc:: drivers/gpu/drm/drm_gpuvm.c <<
484 :doc: Split and Merge <<
485 <<
486 .. _drm_gpuvm_locking: <<
487 <<
488 Locking <<
489 ------- <<
490 <<
491 .. kernel-doc:: drivers/gpu/drm/drm_gpuvm.c <<
492 :doc: Locking <<
493 <<
494 Examples <<
495 -------- <<
496 <<
497 .. kernel-doc:: drivers/gpu/drm/drm_gpuvm.c <<
498 :doc: Examples <<
499 <<
500 DRM GPUVM Function References <<
501 ----------------------------- <<
502 <<
503 .. kernel-doc:: include/drm/drm_gpuvm.h <<
504 :internal: <<
505 <<
506 .. kernel-doc:: drivers/gpu/drm/drm_gpuvm.c <<
507 :export: <<
508 <<
509 DRM Buddy Allocator <<
510 =================== <<
511 <<
512 DRM Buddy Function References <<
513 ----------------------------- <<
514 <<
515 .. kernel-doc:: drivers/gpu/drm/drm_buddy.c <<
516 :export: <<
517 <<
518 DRM Cache Handling and Fast WC memcpy() <<
519 ======================================= <<
520 <<
521 .. kernel-doc:: drivers/gpu/drm/drm_cache.c <<
522 :export: <<
523 <<
524 .. _drm_sync_objects: <<
525 <<
526 DRM Sync Objects <<
527 ================ <<
528 <<
529 .. kernel-doc:: drivers/gpu/drm/drm_syncobj.c <<
530 :doc: Overview <<
531 <<
532 .. kernel-doc:: include/drm/drm_syncobj.h <<
533 :internal: <<
534 <<
535 .. kernel-doc:: drivers/gpu/drm/drm_syncobj.c <<
536 :export: <<
537 <<
538 DRM Execution context <<
539 ===================== <<
540 <<
541 .. kernel-doc:: drivers/gpu/drm/drm_exec.c <<
542 :doc: Overview <<
543 <<
544 .. kernel-doc:: include/drm/drm_exec.h <<
545 :internal: <<
546 <<
547 .. kernel-doc:: drivers/gpu/drm/drm_exec.c <<
548 :export: <<
549 <<
550 GPU Scheduler <<
551 ============= <<
552 <<
553 Overview <<
554 -------- <<
555 <<
556 .. kernel-doc:: drivers/gpu/drm/scheduler/sche <<
557 :doc: Overview <<
558 <<
559 Flow Control <<
560 ------------ <<
561 <<
562 .. kernel-doc:: drivers/gpu/drm/scheduler/sche <<
563 :doc: Flow Control <<
564 <<
565 Scheduler Function References <<
566 ----------------------------- <<
567 <<
568 .. kernel-doc:: include/drm/gpu_scheduler.h <<
569 :internal: 454 :internal:
570 <<
571 .. kernel-doc:: drivers/gpu/drm/scheduler/sche <<
572 :export: <<
573 <<
574 .. kernel-doc:: drivers/gpu/drm/scheduler/sche <<
575 :export: <<
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.