Add kernel-side support for using the zcull hardware in nvidia gpus. zcull aims to improve memory bandwidth by using an early approximate depth test, similar to hierarchical Z on an AMD card. Add a new ioctl that exposes zcull information that has been read from the hardware. Userspace uses each of these parameters either in a heuristic for determining zcull region parameters or in the calculation of a buffer size. It appears the hardware hasn't changed its structure for these values since FERMI_C (circa 2011), so the assumption is that it won't change on us too quickly, and is therefore reasonable to include in UAPI. This bypasses the nvif layer and instead accesses nvkm_gr directly, which mirrors existing usage of nvkm_gr_units(). There is no nvif object for nvkm_gr yet, and adding one is not trivial.
Showing the changes to include/uapi (1 of 4 files). Show all
Unified · Split
| @@ -432,6 +432,69 @@ struct drm_nouveau_exec { · 431 unmodified lines | ||
| 432 | 432 | __u64 push_ptr; |
| 433 | 433 | }; |
| 434 | 434 | |
| 435 | +struct drm_nouveau_get_zcull_info { | |
| 436 | + /** | |
| 437 | + * @width_align_pixels: required alignment for region widths, in pixels | |
| 438 | + * (typically #TPC's * 16). | |
| 439 | + */ | |
| 440 | + __u32 width_align_pixels; | |
| 441 | + /** | |
| 442 | + * @height_align_pixels: required alignment for region heights, in | |
| 443 | + * pixels (typically 32). | |
| 444 | + */ | |
| 445 | + __u32 height_align_pixels; | |
| 446 | + /** | |
| 447 | + * @pixel_squares_by_aliquots: the pixel area covered by an aliquot | |
| 448 | + * (typically #Zcull_banks * 16 * 16). | |
| 449 | + */ | |
| 450 | + __u32 pixel_squares_by_aliquots; | |
| 451 | + /** | |
| 452 | + * @aliquot_total: the total aliquot pool available in hardware | |
| 453 | + */ | |
| 454 | + __u32 aliquot_total; | |
| 455 | + /** | |
| 456 | + * @zcull_region_byte_multiplier: the size of an aliquot in bytes, which | |
| 457 | + * is used for save/restore operations on a region | |
| 458 | + */ | |
| 459 | + __u32 zcull_region_byte_multiplier; | |
| 460 | + /** | |
| 461 | + * @zcull_region_header_size: the region header size in bytes, which is | |
| 462 | + * used for save/restore operations on a region | |
| 463 | + */ | |
| 464 | + __u32 zcull_region_header_size; | |
| 465 | + /** | |
| 466 | + * @zcull_subregion_header_size: the subregion header size in bytes, | |
| 467 | + * which is used for save/restore operations on a region | |
| 468 | + */ | |
| 469 | + __u32 zcull_subregion_header_size; | |
| 470 | + /** | |
| 471 | + * @subregion_count: the total number of subregions the hardware | |
| 472 | + * supports | |
| 473 | + */ | |
| 474 | + __u32 subregion_count; | |
| 475 | + /** | |
| 476 | + * @subregion_width_align_pixels: required alignment for subregion | |
| 477 | + * widths, in pixels (typically #TPC's * 16). | |
| 478 | + */ | |
| 479 | + __u32 subregion_width_align_pixels; | |
| 480 | + /** | |
| 481 | + * @subregion_height_align_pixels: required alignment for subregion | |
| 482 | + * heights, in pixels | |
| 483 | + */ | |
| 484 | + __u32 subregion_height_align_pixels; | |
| 485 | + | |
| 486 | + /** | |
| 487 | + * @ctxsw_size: the size, in bytes, of a zcull context switching region. | |
| 488 | + * Will be zero if the kernel does not support zcull context switching. | |
| 489 | + */ | |
| 490 | + __u32 ctxsw_size; | |
| 491 | + /** | |
| 492 | + * @ctxsw_align: the alignment, in bytes, of a zcull context switching | |
| 493 | + * region | |
| 494 | + */ | |
| 495 | + __u32 ctxsw_align; | |
| 496 | +}; | |
| 497 | + | |
| 435 | 498 | #define DRM_NOUVEAU_GETPARAM 0x00 |
| 436 | 499 | #define DRM_NOUVEAU_SETPARAM 0x01 /* deprecated */ |
| 437 | 500 | #define DRM_NOUVEAU_CHANNEL_ALLOC 0x02 |
| @@ -445,6 +508,7 @@ struct drm_nouveau_exec { · 7 unmodified lines | ||
| 445 | 508 | #define DRM_NOUVEAU_VM_INIT 0x10 |
| 446 | 509 | #define DRM_NOUVEAU_VM_BIND 0x11 |
| 447 | 510 | #define DRM_NOUVEAU_EXEC 0x12 |
| 511 | +#define DRM_NOUVEAU_GET_ZCULL_INFO 0x13 | |
| 448 | 512 | #define DRM_NOUVEAU_GEM_NEW 0x40 |
| 449 | 513 | #define DRM_NOUVEAU_GEM_PUSHBUF 0x41 |
| 450 | 514 | #define DRM_NOUVEAU_GEM_CPU_PREP 0x42 |
| @@ -513,6 +577,8 @@ struct drm_nouveau_svm_bind { · 62 unmodified lines | ||
| 513 | 577 | #define DRM_IOCTL_NOUVEAU_VM_INIT DRM_IOWR(DRM_COMMAND_BASE + DRM_NOUVEAU_VM_INIT, struct drm_nouveau_vm_init) |
| 514 | 578 | #define DRM_IOCTL_NOUVEAU_VM_BIND DRM_IOWR(DRM_COMMAND_BASE + DRM_NOUVEAU_VM_BIND, struct drm_nouveau_vm_bind) |
| 515 | 579 | #define DRM_IOCTL_NOUVEAU_EXEC DRM_IOWR(DRM_COMMAND_BASE + DRM_NOUVEAU_EXEC, struct drm_nouveau_exec) |
| 580 | + | |
| 581 | +#define DRM_IOCTL_NOUVEAU_GET_ZCULL_INFO DRM_IOR (DRM_COMMAND_BASE + DRM_NOUVEAU_GET_ZCULL_INFO, struct drm_nouveau_get_zcull_info) | |
| 516 | 582 | #if defined(__cplusplus) |
| 517 | 583 | } |
| 518 | 584 | #endif |