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56 :internal: !! 37 This section is outdated.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