1===================== 2DRM Memory Management 3===================== 4 5Modern Linux systems require large amount of graphics memory to store 6frame buffers, textures, vertices and other graphics-related data. Given 7the very dynamic nature of many of that data, managing graphics memory 8efficiently is thus crucial for the graphics stack and plays a central 9role in the DRM infrastructure. 10 11The DRM core includes two memory managers, namely Translation Table Manager 12(TTM) and Graphics Execution Manager (GEM). TTM was the first DRM memory 13manager to be developed and tried to be a one-size-fits-them all 14solution. It provides a single userspace API to accommodate the need of 15all hardware, supporting both Unified Memory Architecture (UMA) devices 16and devices with dedicated video RAM (i.e. most discrete video cards). 17This resulted in a large, complex piece of code that turned out to be 18hard to use for driver development. 19 20GEM started as an Intel-sponsored project in reaction to TTM's 21complexity. Its design philosophy is completely different: instead of 22providing a solution to every graphics memory-related problems, GEM 23identified common code between drivers and created a support library to 24share it. GEM has simpler initialization and execution requirements than 25TTM, but has no video RAM management capabilities and is thus limited to 26UMA devices. 27 28The Translation Table Manager (TTM) 29=================================== 30 31.. kernel-doc:: drivers/gpu/drm/ttm/ttm_module.c 32 :doc: TTM 33 34.. kernel-doc:: include/drm/ttm/ttm_caching.h 35 :internal: 36 37TTM 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.c 44 :export: 45 46TTM resource placement reference 47-------------------------------- 48 49.. kernel-doc:: include/drm/ttm/ttm_placement.h 50 :internal: 51 52TTM 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_resource.c 59 :export: 60 61TTM TT object reference 62----------------------- 63 64.. kernel-doc:: include/drm/ttm/ttm_tt.h 65 :internal: 66 67.. kernel-doc:: drivers/gpu/drm/ttm/ttm_tt.c 68 :export: 69 70TTM page pool reference 71----------------------- 72 73.. kernel-doc:: include/drm/ttm/ttm_pool.h 74 :internal: 75 76.. kernel-doc:: drivers/gpu/drm/ttm/ttm_pool.c 77 :export: 78 79The Graphics Execution Manager (GEM) 80==================================== 81 82The GEM design approach has resulted in a memory manager that doesn't 83provide full coverage of all (or even all common) use cases in its 84userspace or kernel API. GEM exposes a set of standard memory-related 85operations to userspace and a set of helper functions to drivers, and 86let drivers implement hardware-specific operations with their own 87private API. 88 89The GEM userspace API is described in the `GEM - the Graphics Execution 90Manager <http://lwn.net/Articles/283798/>`__ article on LWN. While 91slightly outdated, the document provides a good overview of the GEM API 92principles. Buffer allocation and read and write operations, described 93as part of the common GEM API, are currently implemented using 94driver-specific ioctls. 95 96GEM is data-agnostic. It manages abstract buffer objects without knowing 97what individual buffers contain. APIs that require knowledge of buffer 98contents or purpose, such as buffer allocation or synchronization 99primitives, are thus outside of the scope of GEM and must be implemented 100using driver-specific ioctls. 101 102On a fundamental level, GEM involves several operations: 103 104- Memory allocation and freeing 105- Command execution 106- Aperture management at command execution time 107 108Buffer object allocation is relatively straightforward and largely 109provided by Linux's shmem layer, which provides memory to back each 110object. 111 112Device-specific operations, such as command execution, pinning, buffer 113read & write, mapping, and domain ownership transfers are left to 114driver-specific ioctls. 115 116GEM Initialization 117------------------ 118 119Drivers that use GEM must set the DRIVER_GEM bit in the struct 120:c:type:`struct drm_driver <drm_driver>` driver_features 121field. The DRM core will then automatically initialize the GEM core 122before calling the load operation. Behind the scene, this will create a 123DRM Memory Manager object which provides an address space pool for 124object allocation. 125 126In a KMS configuration, drivers need to allocate and initialize a 127command ring buffer following core GEM initialization if required by the 128hardware. UMA devices usually have what is called a "stolen" memory 129region, which provides space for the initial framebuffer and large, 130contiguous memory regions required by the device. This space is 131typically not managed by GEM, and must be initialized separately into 132its own DRM MM object. 133 134GEM Objects Creation 135-------------------- 136 137GEM splits creation of GEM objects and allocation of the memory that 138backs them in two distinct operations. 139 140GEM objects are represented by an instance of struct :c:type:`struct 141drm_gem_object <drm_gem_object>`. Drivers usually need to 142extend GEM objects with private information and thus create a 143driver-specific GEM object structure type that embeds an instance of 144struct :c:type:`struct drm_gem_object <drm_gem_object>`. 145 146To create a GEM object, a driver allocates memory for an instance of its 147specific GEM object type and initializes the embedded struct 148:c:type:`struct drm_gem_object <drm_gem_object>` with a call 149to drm_gem_object_init(). The function takes a pointer 150to the DRM device, a pointer to the GEM object and the buffer object 151size in bytes. 152 153GEM uses shmem to allocate anonymous pageable memory. 154drm_gem_object_init() will create an shmfs file of the 155requested size and store it into the struct :c:type:`struct 156drm_gem_object <drm_gem_object>` filp field. The memory is 157used as either main storage for the object when the graphics hardware 158uses system memory directly or as a backing store otherwise. 159 160Drivers are responsible for the actual physical pages allocation by 161calling shmem_read_mapping_page_gfp() for each page. 162Note that they can decide to allocate pages when initializing the GEM 163object, or to delay allocation until the memory is needed (for instance 164when a page fault occurs as a result of a userspace memory access or 165when the driver needs to start a DMA transfer involving the memory). 166 167Anonymous pageable memory allocation is not always desired, for instance 168when the hardware requires physically contiguous system memory as is 169often the case in embedded devices. Drivers can create GEM objects with 170no shmfs backing (called private GEM objects) by initializing them with a call 171to drm_gem_private_object_init() instead of drm_gem_object_init(). Storage for 172private GEM objects must be managed by drivers. 173 174GEM Objects Lifetime 175-------------------- 176 177All GEM objects are reference-counted by the GEM core. References can be 178acquired and release by calling drm_gem_object_get() and drm_gem_object_put() 179respectively. 180 181When the last reference to a GEM object is released the GEM core calls 182the :c:type:`struct drm_gem_object_funcs <gem_object_funcs>` free 183operation. That operation is mandatory for GEM-enabled drivers and must 184free the GEM object and all associated resources. 185 186void (\*free) (struct drm_gem_object \*obj); Drivers are 187responsible for freeing all GEM object resources. This includes the 188resources created by the GEM core, which need to be released with 189drm_gem_object_release(). 190 191GEM Objects Naming 192------------------ 193 194Communication between userspace and the kernel refers to GEM objects 195using local handles, global names or, more recently, file descriptors. 196All of those are 32-bit integer values; the usual Linux kernel limits 197apply to the file descriptors. 198 199GEM handles are local to a DRM file. Applications get a handle to a GEM 200object through a driver-specific ioctl, and can use that handle to refer 201to the GEM object in other standard or driver-specific ioctls. Closing a 202DRM file handle frees all its GEM handles and dereferences the 203associated GEM objects. 204 205To create a handle for a GEM object drivers call drm_gem_handle_create(). The 206function takes a pointer to the DRM file and the GEM object and returns a 207locally unique handle. When the handle is no longer needed drivers delete it 208with a call to drm_gem_handle_delete(). Finally the GEM object associated with a 209handle can be retrieved by a call to drm_gem_object_lookup(). 210 211Handles don't take ownership of GEM objects, they only take a reference 212to the object that will be dropped when the handle is destroyed. To 213avoid leaking GEM objects, drivers must make sure they drop the 214reference(s) they own (such as the initial reference taken at object 215creation time) as appropriate, without any special consideration for the 216handle. For example, in the particular case of combined GEM object and 217handle creation in the implementation of the dumb_create operation, 218drivers must drop the initial reference to the GEM object before 219returning the handle. 220 221GEM names are similar in purpose to handles but are not local to DRM 222files. They can be passed between processes to reference a GEM object 223globally. Names can't be used directly to refer to objects in the DRM 224API, applications must convert handles to names and names to handles 225using the DRM_IOCTL_GEM_FLINK and DRM_IOCTL_GEM_OPEN ioctls 226respectively. The conversion is handled by the DRM core without any 227driver-specific support. 228 229GEM also supports buffer sharing with dma-buf file descriptors through 230PRIME. GEM-based drivers must use the provided helpers functions to 231implement the exporting and importing correctly. See ?. Since sharing 232file descriptors is inherently more secure than the easily guessable and 233global GEM names it is the preferred buffer sharing mechanism. Sharing 234buffers through GEM names is only supported for legacy userspace. 235Furthermore PRIME also allows cross-device buffer sharing since it is 236based on dma-bufs. 237 238GEM Objects Mapping 239------------------- 240 241Because mapping operations are fairly heavyweight GEM favours 242read/write-like access to buffers, implemented through driver-specific 243ioctls, over mapping buffers to userspace. However, when random access 244to the buffer is needed (to perform software rendering for instance), 245direct access to the object can be more efficient. 246 247The mmap system call can't be used directly to map GEM objects, as they 248don't have their own file handle. Two alternative methods currently 249co-exist to map GEM objects to userspace. The first method uses a 250driver-specific ioctl to perform the mapping operation, calling 251do_mmap() under the hood. This is often considered 252dubious, seems to be discouraged for new GEM-enabled drivers, and will 253thus not be described here. 254 255The second method uses the mmap system call on the DRM file handle. void 256\*mmap(void \*addr, size_t length, int prot, int flags, int fd, off_t 257offset); DRM identifies the GEM object to be mapped by a fake offset 258passed through the mmap offset argument. Prior to being mapped, a GEM 259object must thus be associated with a fake offset. To do so, drivers 260must call drm_gem_create_mmap_offset() on the object. 261 262Once allocated, the fake offset value must be passed to the application 263in a driver-specific way and can then be used as the mmap offset 264argument. 265 266The GEM core provides a helper method drm_gem_mmap() to 267handle object mapping. The method can be set directly as the mmap file 268operation handler. It will look up the GEM object based on the offset 269value and set the VMA operations to the :c:type:`struct drm_driver 270<drm_driver>` gem_vm_ops field. Note that drm_gem_mmap() doesn't map memory to 271userspace, but relies on the driver-provided fault handler to map pages 272individually. 273 274To use drm_gem_mmap(), drivers must fill the struct :c:type:`struct drm_driver 275<drm_driver>` gem_vm_ops field with a pointer to VM operations. 276 277The VM operations is a :c:type:`struct vm_operations_struct <vm_operations_struct>` 278made up of several fields, the more interesting ones being: 279 280.. code-block:: c 281 282 struct vm_operations_struct { 283 void (*open)(struct vm_area_struct * area); 284 void (*close)(struct vm_area_struct * area); 285 vm_fault_t (*fault)(struct vm_fault *vmf); 286 }; 287 288 289The open and close operations must update the GEM object reference 290count. Drivers can use the drm_gem_vm_open() and drm_gem_vm_close() helper 291functions directly as open and close handlers. 292 293The fault operation handler is responsible for mapping individual pages 294to userspace when a page fault occurs. Depending on the memory 295allocation scheme, drivers can allocate pages at fault time, or can 296decide to allocate memory for the GEM object at the time the object is 297created. 298 299Drivers that want to map the GEM object upfront instead of handling page 300faults can implement their own mmap file operation handler. 301 302For platforms without MMU the GEM core provides a helper method 303drm_gem_dma_get_unmapped_area(). The mmap() routines will call this to get a 304proposed address for the mapping. 305 306To use drm_gem_dma_get_unmapped_area(), drivers must fill the struct 307:c:type:`struct file_operations <file_operations>` get_unmapped_area field with 308a pointer on drm_gem_dma_get_unmapped_area(). 309 310More detailed information about get_unmapped_area can be found in 311Documentation/admin-guide/mm/nommu-mmap.rst 312 313Memory Coherency 314---------------- 315 316When mapped to the device or used in a command buffer, backing pages for 317an object are flushed to memory and marked write combined so as to be 318coherent with the GPU. Likewise, if the CPU accesses an object after the 319GPU has finished rendering to the object, then the object must be made 320coherent with the CPU's view of memory, usually involving GPU cache 321flushing of various kinds. This core CPU<->GPU coherency management is 322provided by a device-specific ioctl, which evaluates an object's current 323domain and performs any necessary flushing or synchronization to put the 324object into the desired coherency domain (note that the object may be 325busy, i.e. an active render target; in that case, setting the domain 326blocks the client and waits for rendering to complete before performing 327any necessary flushing operations). 328 329Command Execution 330----------------- 331 332Perhaps the most important GEM function for GPU devices is providing a 333command execution interface to clients. Client programs construct 334command buffers containing references to previously allocated memory 335objects, and then submit them to GEM. At that point, GEM takes care to 336bind all the objects into the GTT, execute the buffer, and provide 337necessary synchronization between clients accessing the same buffers. 338This often involves evicting some objects from the GTT and re-binding 339others (a fairly expensive operation), and providing relocation support 340which hides fixed GTT offsets from clients. Clients must take care not 341to submit command buffers that reference more objects than can fit in 342the GTT; otherwise, GEM will reject them and no rendering will occur. 343Similarly, if several objects in the buffer require fence registers to 344be allocated for correct rendering (e.g. 2D blits on pre-965 chips), 345care must be taken not to require more fence registers than are 346available to the client. Such resource management should be abstracted 347from the client in libdrm. 348 349GEM Function Reference 350---------------------- 351 352.. kernel-doc:: include/drm/drm_gem.h 353 :internal: 354 355.. kernel-doc:: drivers/gpu/drm/drm_gem.c 356 :export: 357 358GEM DMA Helper Functions Reference 359---------------------------------- 360 361.. kernel-doc:: drivers/gpu/drm/drm_gem_dma_helper.c 362 :doc: dma helpers 363 364.. kernel-doc:: include/drm/drm_gem_dma_helper.h 365 :internal: 366 367.. kernel-doc:: drivers/gpu/drm/drm_gem_dma_helper.c 368 :export: 369 370GEM SHMEM Helper Function Reference 371----------------------------------- 372 373.. kernel-doc:: drivers/gpu/drm/drm_gem_shmem_helper.c 374 :doc: overview 375 376.. kernel-doc:: include/drm/drm_gem_shmem_helper.h 377 :internal: 378 379.. kernel-doc:: drivers/gpu/drm/drm_gem_shmem_helper.c 380 :export: 381 382GEM VRAM Helper Functions Reference 383----------------------------------- 384 385.. kernel-doc:: drivers/gpu/drm/drm_gem_vram_helper.c 386 :doc: overview 387 388.. kernel-doc:: include/drm/drm_gem_vram_helper.h 389 :internal: 390 391.. kernel-doc:: drivers/gpu/drm/drm_gem_vram_helper.c 392 :export: 393 394GEM TTM Helper Functions Reference 395----------------------------------- 396 397.. kernel-doc:: drivers/gpu/drm/drm_gem_ttm_helper.c 398 :doc: overview 399 400.. kernel-doc:: drivers/gpu/drm/drm_gem_ttm_helper.c 401 :export: 402 403VMA Offset Manager 404================== 405 406.. kernel-doc:: drivers/gpu/drm/drm_vma_manager.c 407 :doc: vma offset manager 408 409.. kernel-doc:: include/drm/drm_vma_manager.h 410 :internal: 411 412.. kernel-doc:: drivers/gpu/drm/drm_vma_manager.c 413 :export: 414 415.. _prime_buffer_sharing: 416 417PRIME Buffer Sharing 418==================== 419 420PRIME is the cross device buffer sharing framework in drm, originally 421created for the OPTIMUS range of multi-gpu platforms. To userspace PRIME 422buffers are dma-buf based file descriptors. 423 424Overview and Lifetime Rules 425--------------------------- 426 427.. kernel-doc:: drivers/gpu/drm/drm_prime.c 428 :doc: overview and lifetime rules 429 430PRIME Helper Functions 431---------------------- 432 433.. kernel-doc:: drivers/gpu/drm/drm_prime.c 434 :doc: PRIME Helpers 435 436PRIME Function References 437------------------------- 438 439.. kernel-doc:: include/drm/drm_prime.h 440 :internal: 441 442.. kernel-doc:: drivers/gpu/drm/drm_prime.c 443 :export: 444 445DRM MM Range Allocator 446====================== 447 448Overview 449-------- 450 451.. kernel-doc:: drivers/gpu/drm/drm_mm.c 452 :doc: Overview 453 454LRU Scan/Eviction Support 455------------------------- 456 457.. kernel-doc:: drivers/gpu/drm/drm_mm.c 458 :doc: lru scan roster 459 460DRM MM Range Allocator Function References 461------------------------------------------ 462 463.. kernel-doc:: include/drm/drm_mm.h 464 :internal: 465 466.. kernel-doc:: drivers/gpu/drm/drm_mm.c 467 :export: 468 469.. _drm_gpuvm: 470 471DRM GPUVM 472========= 473 474Overview 475-------- 476 477.. kernel-doc:: drivers/gpu/drm/drm_gpuvm.c 478 :doc: Overview 479 480Split and Merge 481--------------- 482 483.. kernel-doc:: drivers/gpu/drm/drm_gpuvm.c 484 :doc: Split and Merge 485 486.. _drm_gpuvm_locking: 487 488Locking 489------- 490 491.. kernel-doc:: drivers/gpu/drm/drm_gpuvm.c 492 :doc: Locking 493 494Examples 495-------- 496 497.. kernel-doc:: drivers/gpu/drm/drm_gpuvm.c 498 :doc: Examples 499 500DRM 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 509DRM Buddy Allocator 510=================== 511 512DRM Buddy Function References 513----------------------------- 514 515.. kernel-doc:: drivers/gpu/drm/drm_buddy.c 516 :export: 517 518DRM 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 526DRM 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 538DRM 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 550GPU Scheduler 551============= 552 553Overview 554-------- 555 556.. kernel-doc:: drivers/gpu/drm/scheduler/sched_main.c 557 :doc: Overview 558 559Flow Control 560------------ 561 562.. kernel-doc:: drivers/gpu/drm/scheduler/sched_main.c 563 :doc: Flow Control 564 565Scheduler Function References 566----------------------------- 567 568.. kernel-doc:: include/drm/gpu_scheduler.h 569 :internal: 570 571.. kernel-doc:: drivers/gpu/drm/scheduler/sched_main.c 572 :export: 573 574.. kernel-doc:: drivers/gpu/drm/scheduler/sched_entity.c 575 :export: 576