diff options
Diffstat (limited to 'drivers/gpu/drm/nouveau/nouveau_bo.c')
| -rw-r--r-- | drivers/gpu/drm/nouveau/nouveau_bo.c | 297 |
1 files changed, 135 insertions, 162 deletions
diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.c b/drivers/gpu/drm/nouveau/nouveau_bo.c index 69d7b1d0b9d..b6dc85c614b 100644 --- a/drivers/gpu/drm/nouveau/nouveau_bo.c +++ b/drivers/gpu/drm/nouveau/nouveau_bo.c @@ -28,6 +28,7 @@ */ #include <core/engine.h> +#include <linux/swiotlb.h> #include <subdev/fb.h> #include <subdev/vm.h> @@ -97,12 +98,7 @@ nv10_bo_put_tile_region(struct drm_device *dev, struct nouveau_drm_tile *tile, if (tile) { spin_lock(&drm->tile.lock); - if (fence) { - /* Mark it as pending. */ - tile->fence = fence; - nouveau_fence_ref(fence); - } - + tile->fence = nouveau_fence_ref(fence); tile->used = false; spin_unlock(&drm->tile.lock); } @@ -145,8 +141,9 @@ nouveau_bo_del_ttm(struct ttm_buffer_object *bo) struct drm_device *dev = drm->dev; struct nouveau_bo *nvbo = nouveau_bo(bo); - if (unlikely(nvbo->gem)) + if (unlikely(nvbo->gem.filp)) DRM_ERROR("bo %p still attached to GEM object\n", bo); + WARN_ON(nvbo->pin_refcnt > 0); nv10_bo_put_tile_region(dev, nvbo->tile, NULL); kfree(nvbo); } @@ -196,6 +193,17 @@ nouveau_bo_new(struct drm_device *dev, int size, int align, size_t acc_size; int ret; int type = ttm_bo_type_device; + int lpg_shift = 12; + int max_size; + + if (drm->client.base.vm) + lpg_shift = drm->client.base.vm->vmm->lpg_shift; + max_size = INT_MAX & ~((1 << lpg_shift) - 1); + + if (size <= 0 || size > max_size) { + nv_warn(drm, "skipped size %x\n", (u32)size); + return -EINVAL; + } if (sg) type = ttm_bo_type_sg; @@ -254,9 +262,10 @@ set_placement_range(struct nouveau_bo *nvbo, uint32_t type) { struct nouveau_drm *drm = nouveau_bdev(nvbo->bo.bdev); struct nouveau_fb *pfb = nouveau_fb(drm->device); - u32 vram_pages = pfb->ram.size >> PAGE_SHIFT; + u32 vram_pages = pfb->ram->size >> PAGE_SHIFT; - if (nv_device(drm->device)->card_type == NV_10 && + if ((nv_device(drm->device)->card_type == NV_10 || + nv_device(drm->device)->card_type == NV_11) && nvbo->tile_mode && (type & TTM_PL_FLAG_VRAM) && nvbo->bo.mem.num_pages < vram_pages / 4) { /* @@ -300,17 +309,18 @@ nouveau_bo_pin(struct nouveau_bo *nvbo, uint32_t memtype) struct ttm_buffer_object *bo = &nvbo->bo; int ret; + ret = ttm_bo_reserve(bo, false, false, false, 0); + if (ret) + goto out; + if (nvbo->pin_refcnt && !(memtype & (1 << bo->mem.mem_type))) { NV_ERROR(drm, "bo %p pinned elsewhere: 0x%08x vs 0x%08x\n", bo, 1 << bo->mem.mem_type, memtype); - return -EINVAL; + ret = -EINVAL; + goto out; } if (nvbo->pin_refcnt++) - return 0; - - ret = ttm_bo_reserve(bo, false, false, false, 0); - if (ret) goto out; nouveau_bo_placement_set(nvbo, memtype, 0); @@ -328,10 +338,8 @@ nouveau_bo_pin(struct nouveau_bo *nvbo, uint32_t memtype) break; } } - ttm_bo_unreserve(bo); out: - if (unlikely(ret)) - nvbo->pin_refcnt--; + ttm_bo_unreserve(bo); return ret; } @@ -340,15 +348,17 @@ nouveau_bo_unpin(struct nouveau_bo *nvbo) { struct nouveau_drm *drm = nouveau_bdev(nvbo->bo.bdev); struct ttm_buffer_object *bo = &nvbo->bo; - int ret; - - if (--nvbo->pin_refcnt) - return 0; + int ret, ref; ret = ttm_bo_reserve(bo, false, false, false, 0); if (ret) return ret; + ref = --nvbo->pin_refcnt; + WARN_ON_ONCE(ref < 0); + if (ref) + goto out; + nouveau_bo_placement_set(nvbo, bo->mem.placement, 0); ret = nouveau_bo_validate(nvbo, false, false); @@ -365,6 +375,7 @@ nouveau_bo_unpin(struct nouveau_bo *nvbo) } } +out: ttm_bo_unreserve(bo); return ret; } @@ -549,35 +560,13 @@ nouveau_bo_evict_flags(struct ttm_buffer_object *bo, struct ttm_placement *pl) } -/* GPU-assisted copy using NV_MEMORY_TO_MEMORY_FORMAT, can access - * TTM_PL_{VRAM,TT} directly. - */ - -static int -nouveau_bo_move_accel_cleanup(struct nouveau_channel *chan, - struct nouveau_bo *nvbo, bool evict, - bool no_wait_gpu, struct ttm_mem_reg *new_mem) -{ - struct nouveau_fence *fence = NULL; - int ret; - - ret = nouveau_fence_new(chan, &fence); - if (ret) - return ret; - - ret = ttm_bo_move_accel_cleanup(&nvbo->bo, fence, evict, - no_wait_gpu, new_mem); - nouveau_fence_unref(&fence); - return ret; -} - static int nve0_bo_move_init(struct nouveau_channel *chan, u32 handle) { int ret = RING_SPACE(chan, 2); if (ret == 0) { BEGIN_NVC0(chan, NvSubCopy, 0x0000, 1); - OUT_RING (chan, handle); + OUT_RING (chan, handle & 0x0000ffff); FIRE_RING (chan); } return ret; @@ -787,25 +776,25 @@ nv50_bo_move_m2mf(struct nouveau_channel *chan, struct ttm_buffer_object *bo, struct ttm_mem_reg *old_mem, struct ttm_mem_reg *new_mem) { struct nouveau_mem *node = old_mem->mm_node; - struct nouveau_bo *nvbo = nouveau_bo(bo); u64 length = (new_mem->num_pages << PAGE_SHIFT); u64 src_offset = node->vma[0].offset; u64 dst_offset = node->vma[1].offset; + int src_tiled = !!node->memtype; + int dst_tiled = !!((struct nouveau_mem *)new_mem->mm_node)->memtype; int ret; while (length) { u32 amount, stride, height; + ret = RING_SPACE(chan, 18 + 6 * (src_tiled + dst_tiled)); + if (ret) + return ret; + amount = min(length, (u64)(4 * 1024 * 1024)); stride = 16 * 4; height = amount / stride; - if (new_mem->mem_type == TTM_PL_VRAM && - nouveau_bo_tile_layout(nvbo)) { - ret = RING_SPACE(chan, 8); - if (ret) - return ret; - + if (src_tiled) { BEGIN_NV04(chan, NvSubCopy, 0x0200, 7); OUT_RING (chan, 0); OUT_RING (chan, 0); @@ -815,19 +804,10 @@ nv50_bo_move_m2mf(struct nouveau_channel *chan, struct ttm_buffer_object *bo, OUT_RING (chan, 0); OUT_RING (chan, 0); } else { - ret = RING_SPACE(chan, 2); - if (ret) - return ret; - BEGIN_NV04(chan, NvSubCopy, 0x0200, 1); OUT_RING (chan, 1); } - if (old_mem->mem_type == TTM_PL_VRAM && - nouveau_bo_tile_layout(nvbo)) { - ret = RING_SPACE(chan, 8); - if (ret) - return ret; - + if (dst_tiled) { BEGIN_NV04(chan, NvSubCopy, 0x021c, 7); OUT_RING (chan, 0); OUT_RING (chan, 0); @@ -837,18 +817,10 @@ nv50_bo_move_m2mf(struct nouveau_channel *chan, struct ttm_buffer_object *bo, OUT_RING (chan, 0); OUT_RING (chan, 0); } else { - ret = RING_SPACE(chan, 2); - if (ret) - return ret; - BEGIN_NV04(chan, NvSubCopy, 0x021c, 1); OUT_RING (chan, 1); } - ret = RING_SPACE(chan, 14); - if (ret) - return ret; - BEGIN_NV04(chan, NvSubCopy, 0x0238, 2); OUT_RING (chan, upper_32_bits(src_offset)); OUT_RING (chan, upper_32_bits(dst_offset)); @@ -942,23 +914,28 @@ nv04_bo_move_m2mf(struct nouveau_channel *chan, struct ttm_buffer_object *bo, } static int -nouveau_vma_getmap(struct nouveau_channel *chan, struct nouveau_bo *nvbo, - struct ttm_mem_reg *mem, struct nouveau_vma *vma) +nouveau_bo_move_prep(struct nouveau_drm *drm, struct ttm_buffer_object *bo, + struct ttm_mem_reg *mem) { - struct nouveau_mem *node = mem->mm_node; + struct nouveau_mem *old_node = bo->mem.mm_node; + struct nouveau_mem *new_node = mem->mm_node; + u64 size = (u64)mem->num_pages << PAGE_SHIFT; int ret; - ret = nouveau_vm_get(nv_client(chan->cli)->vm, mem->num_pages << - PAGE_SHIFT, node->page_shift, - NV_MEM_ACCESS_RW, vma); + ret = nouveau_vm_get(nv_client(drm)->vm, size, old_node->page_shift, + NV_MEM_ACCESS_RW, &old_node->vma[0]); if (ret) return ret; - if (mem->mem_type == TTM_PL_VRAM) - nouveau_vm_map(vma, node); - else - nouveau_vm_map_sg(vma, 0, mem->num_pages << PAGE_SHIFT, node); + ret = nouveau_vm_get(nv_client(drm)->vm, size, new_node->page_shift, + NV_MEM_ACCESS_RW, &old_node->vma[1]); + if (ret) { + nouveau_vm_put(&old_node->vma[0]); + return ret; + } + nouveau_vm_map(&old_node->vma[0], old_node); + nouveau_vm_map(&old_node->vma[1], new_node); return 0; } @@ -967,36 +944,35 @@ nouveau_bo_move_m2mf(struct ttm_buffer_object *bo, int evict, bool intr, bool no_wait_gpu, struct ttm_mem_reg *new_mem) { struct nouveau_drm *drm = nouveau_bdev(bo->bdev); - struct nouveau_channel *chan = chan = drm->channel; - struct nouveau_bo *nvbo = nouveau_bo(bo); - struct ttm_mem_reg *old_mem = &bo->mem; + struct nouveau_channel *chan = drm->ttm.chan; + struct nouveau_fence *fence; int ret; - mutex_lock(&chan->cli->mutex); - /* create temporary vmas for the transfer and attach them to the * old nouveau_mem node, these will get cleaned up after ttm has * destroyed the ttm_mem_reg */ if (nv_device(drm->device)->card_type >= NV_50) { - struct nouveau_mem *node = old_mem->mm_node; - - ret = nouveau_vma_getmap(chan, nvbo, old_mem, &node->vma[0]); + ret = nouveau_bo_move_prep(drm, bo, new_mem); if (ret) - goto out; - - ret = nouveau_vma_getmap(chan, nvbo, new_mem, &node->vma[1]); - if (ret) - goto out; + return ret; } - ret = drm->ttm.move(chan, bo, &bo->mem, new_mem); + mutex_lock_nested(&chan->cli->mutex, SINGLE_DEPTH_NESTING); + ret = nouveau_fence_sync(bo->sync_obj, chan); if (ret == 0) { - ret = nouveau_bo_move_accel_cleanup(chan, nvbo, evict, - no_wait_gpu, new_mem); + ret = drm->ttm.move(chan, bo, &bo->mem, new_mem); + if (ret == 0) { + ret = nouveau_fence_new(chan, false, &fence); + if (ret == 0) { + ret = ttm_bo_move_accel_cleanup(bo, fence, + evict, + no_wait_gpu, + new_mem); + nouveau_fence_unref(&fence); + } + } } - -out: mutex_unlock(&chan->cli->mutex); return ret; } @@ -1013,7 +989,7 @@ nouveau_bo_move_init(struct nouveau_drm *drm) struct ttm_mem_reg *, struct ttm_mem_reg *); int (*init)(struct nouveau_channel *, u32 handle); } _methods[] = { - { "COPY", 0, 0xa0b5, nve0_bo_move_copy, nve0_bo_move_init }, + { "COPY", 4, 0xa0b5, nve0_bo_move_copy, nve0_bo_move_init }, { "GRCE", 0, 0xa0b5, nve0_bo_move_copy, nvc0_bo_move_init }, { "COPY1", 5, 0x90b8, nvc0_bo_move_copy, nvc0_bo_move_init }, { "COPY0", 4, 0x90b5, nvc0_bo_move_copy, nvc0_bo_move_init }, @@ -1033,7 +1009,7 @@ nouveau_bo_move_init(struct nouveau_drm *drm) struct nouveau_channel *chan; u32 handle = (mthd->engine << 16) | mthd->oclass; - if (mthd->init == nve0_bo_move_init) + if (mthd->engine) chan = drm->cechan; else chan = drm->channel; @@ -1051,6 +1027,7 @@ nouveau_bo_move_init(struct nouveau_drm *drm) } drm->ttm.move = mthd->exec; + drm->ttm.chan = chan; name = mthd->name; break; } @@ -1135,19 +1112,10 @@ nouveau_bo_move_ntfy(struct ttm_buffer_object *bo, struct ttm_mem_reg *new_mem) return; list_for_each_entry(vma, &nvbo->vma_list, head) { - if (new_mem && new_mem->mem_type == TTM_PL_VRAM) { + if (new_mem && new_mem->mem_type != TTM_PL_SYSTEM && + (new_mem->mem_type == TTM_PL_VRAM || + nvbo->page_shift != vma->vm->vmm->lpg_shift)) { nouveau_vm_map(vma, new_mem->mm_node); - } else - if (new_mem && new_mem->mem_type == TTM_PL_TT && - nvbo->page_shift == vma->vm->vmm->spg_shift) { - if (((struct nouveau_mem *)new_mem->mm_node)->sg) - nouveau_vm_map_sg_table(vma, 0, new_mem-> - num_pages << PAGE_SHIFT, - new_mem->mm_node); - else - nouveau_vm_map_sg(vma, 0, new_mem-> - num_pages << PAGE_SHIFT, - new_mem->mm_node); } else { nouveau_vm_unmap(vma); } @@ -1212,28 +1180,27 @@ nouveau_bo_move(struct ttm_buffer_object *bo, bool evict, bool intr, goto out; } - /* CPU copy if we have no accelerated method available */ - if (!drm->ttm.move) { - ret = ttm_bo_move_memcpy(bo, evict, no_wait_gpu, new_mem); - goto out; - } - /* Hardware assisted copy. */ - if (new_mem->mem_type == TTM_PL_SYSTEM) - ret = nouveau_bo_move_flipd(bo, evict, intr, - no_wait_gpu, new_mem); - else if (old_mem->mem_type == TTM_PL_SYSTEM) - ret = nouveau_bo_move_flips(bo, evict, intr, - no_wait_gpu, new_mem); - else - ret = nouveau_bo_move_m2mf(bo, evict, intr, - no_wait_gpu, new_mem); - - if (!ret) - goto out; + if (drm->ttm.move) { + if (new_mem->mem_type == TTM_PL_SYSTEM) + ret = nouveau_bo_move_flipd(bo, evict, intr, + no_wait_gpu, new_mem); + else if (old_mem->mem_type == TTM_PL_SYSTEM) + ret = nouveau_bo_move_flips(bo, evict, intr, + no_wait_gpu, new_mem); + else + ret = nouveau_bo_move_m2mf(bo, evict, intr, + no_wait_gpu, new_mem); + if (!ret) + goto out; + } /* Fallback to software copy. */ - ret = ttm_bo_move_memcpy(bo, evict, no_wait_gpu, new_mem); + spin_lock(&bo->bdev->fence_lock); + ret = ttm_bo_wait(bo, true, intr, no_wait_gpu); + spin_unlock(&bo->bdev->fence_lock); + if (ret == 0) + ret = ttm_bo_move_memcpy(bo, evict, no_wait_gpu, new_mem); out: if (nv_device(drm->device)->card_type < NV_50) { @@ -1249,7 +1216,9 @@ out: static int nouveau_bo_verify_access(struct ttm_buffer_object *bo, struct file *filp) { - return 0; + struct nouveau_bo *nvbo = nouveau_bo(bo); + + return drm_vma_node_verify_access(&nvbo->gem.vma_node, filp); } static int @@ -1257,6 +1226,7 @@ nouveau_ttm_io_mem_reserve(struct ttm_bo_device *bdev, struct ttm_mem_reg *mem) { struct ttm_mem_type_manager *man = &bdev->man[mem->mem_type]; struct nouveau_drm *drm = nouveau_bdev(bdev); + struct nouveau_mem *node = mem->mm_node; struct drm_device *dev = drm->dev; int ret; @@ -1279,14 +1249,16 @@ nouveau_ttm_io_mem_reserve(struct ttm_bo_device *bdev, struct ttm_mem_reg *mem) mem->bus.is_iomem = !dev->agp->cant_use_aperture; } #endif - break; + if (nv_device(drm->device)->card_type < NV_50 || !node->memtype) + /* untiled */ + break; + /* fallthrough, tiled memory */ case TTM_PL_VRAM: mem->bus.offset = mem->start << PAGE_SHIFT; - mem->bus.base = pci_resource_start(dev->pdev, 1); + mem->bus.base = nv_device_resource_start(nouveau_dev(dev), 1); mem->bus.is_iomem = true; if (nv_device(drm->device)->card_type >= NV_50) { struct nouveau_bar *bar = nouveau_bar(drm->device); - struct nouveau_mem *node = mem->mm_node; ret = bar->umap(bar, node, NV_MEM_ACCESS_RW, &node->bar_vma); @@ -1321,7 +1293,8 @@ nouveau_ttm_fault_reserve_notify(struct ttm_buffer_object *bo) struct nouveau_drm *drm = nouveau_bdev(bo->bdev); struct nouveau_bo *nvbo = nouveau_bo(bo); struct nouveau_device *device = nv_device(drm->device); - u32 mappable = pci_resource_len(device->pdev, 1) >> PAGE_SHIFT; + u32 mappable = nv_device_resource_len(device, 1) >> PAGE_SHIFT; + int ret; /* as long as the bo isn't in vram, and isn't tiled, we've got * nothing to do here. @@ -1330,10 +1303,20 @@ nouveau_ttm_fault_reserve_notify(struct ttm_buffer_object *bo) if (nv_device(drm->device)->card_type < NV_50 || !nouveau_bo_tile_layout(nvbo)) return 0; + + if (bo->mem.mem_type == TTM_PL_SYSTEM) { + nouveau_bo_placement_set(nvbo, TTM_PL_TT, 0); + + ret = nouveau_bo_validate(nvbo, false, false); + if (ret) + return ret; + } + return 0; } /* make sure bo is in mappable vram */ - if (bo->mem.start + bo->mem.num_pages < mappable) + if (nv_device(drm->device)->card_type >= NV_50 || + bo->mem.start + bo->mem.num_pages < mappable) return 0; @@ -1348,6 +1331,7 @@ nouveau_ttm_tt_populate(struct ttm_tt *ttm) { struct ttm_dma_tt *ttm_dma = (void *)ttm; struct nouveau_drm *drm; + struct nouveau_device *device; struct drm_device *dev; unsigned i; int r; @@ -1365,6 +1349,7 @@ nouveau_ttm_tt_populate(struct ttm_tt *ttm) } drm = nouveau_bdev(ttm->bdev); + device = nv_device(drm->device); dev = drm->dev; #if __OS_HAS_AGP @@ -1385,13 +1370,12 @@ nouveau_ttm_tt_populate(struct ttm_tt *ttm) } for (i = 0; i < ttm->num_pages; i++) { - ttm_dma->dma_address[i] = pci_map_page(dev->pdev, ttm->pages[i], - 0, PAGE_SIZE, - PCI_DMA_BIDIRECTIONAL); - if (pci_dma_mapping_error(dev->pdev, ttm_dma->dma_address[i])) { + ttm_dma->dma_address[i] = nv_device_map_page(device, + ttm->pages[i]); + if (!ttm_dma->dma_address[i]) { while (--i) { - pci_unmap_page(dev->pdev, ttm_dma->dma_address[i], - PAGE_SIZE, PCI_DMA_BIDIRECTIONAL); + nv_device_unmap_page(device, + ttm_dma->dma_address[i]); ttm_dma->dma_address[i] = 0; } ttm_pool_unpopulate(ttm); @@ -1406,6 +1390,7 @@ nouveau_ttm_tt_unpopulate(struct ttm_tt *ttm) { struct ttm_dma_tt *ttm_dma = (void *)ttm; struct nouveau_drm *drm; + struct nouveau_device *device; struct drm_device *dev; unsigned i; bool slave = !!(ttm->page_flags & TTM_PAGE_FLAG_SG); @@ -1414,6 +1399,7 @@ nouveau_ttm_tt_unpopulate(struct ttm_tt *ttm) return; drm = nouveau_bdev(ttm->bdev); + device = nv_device(drm->device); dev = drm->dev; #if __OS_HAS_AGP @@ -1432,8 +1418,7 @@ nouveau_ttm_tt_unpopulate(struct ttm_tt *ttm) for (i = 0; i < ttm->num_pages; i++) { if (ttm_dma->dma_address[i]) { - pci_unmap_page(dev->pdev, ttm_dma->dma_address[i], - PAGE_SIZE, PCI_DMA_BIDIRECTIONAL); + nv_device_unmap_page(device, ttm_dma->dma_address[i]); } } @@ -1443,14 +1428,12 @@ nouveau_ttm_tt_unpopulate(struct ttm_tt *ttm) void nouveau_bo_fence(struct nouveau_bo *nvbo, struct nouveau_fence *fence) { + struct nouveau_fence *new_fence = nouveau_fence_ref(fence); struct nouveau_fence *old_fence = NULL; - if (likely(fence)) - nouveau_fence_ref(fence); - spin_lock(&nvbo->bo.bdev->fence_lock); old_fence = nvbo->bo.sync_obj; - nvbo->bo.sync_obj = fence; + nvbo->bo.sync_obj = new_fence; spin_unlock(&nvbo->bo.bdev->fence_lock); nouveau_fence_unref(&old_fence); @@ -1523,7 +1506,6 @@ nouveau_bo_vma_add(struct nouveau_bo *nvbo, struct nouveau_vm *vm, struct nouveau_vma *vma) { const u32 size = nvbo->bo.mem.num_pages << PAGE_SHIFT; - struct nouveau_mem *node = nvbo->bo.mem.mm_node; int ret; ret = nouveau_vm_get(vm, size, nvbo->page_shift, @@ -1531,14 +1513,10 @@ nouveau_bo_vma_add(struct nouveau_bo *nvbo, struct nouveau_vm *vm, if (ret) return ret; - if (nvbo->bo.mem.mem_type == TTM_PL_VRAM) + if ( nvbo->bo.mem.mem_type != TTM_PL_SYSTEM && + (nvbo->bo.mem.mem_type == TTM_PL_VRAM || + nvbo->page_shift != vma->vm->vmm->lpg_shift)) nouveau_vm_map(vma, nvbo->bo.mem.mm_node); - else if (nvbo->bo.mem.mem_type == TTM_PL_TT) { - if (node->sg) - nouveau_vm_map_sg_table(vma, 0, size, node); - else - nouveau_vm_map_sg(vma, 0, size, node); - } list_add_tail(&vma->head, &nvbo->vma_list); vma->refcount = 1; @@ -1549,13 +1527,8 @@ void nouveau_bo_vma_del(struct nouveau_bo *nvbo, struct nouveau_vma *vma) { if (vma->node) { - if (nvbo->bo.mem.mem_type != TTM_PL_SYSTEM) { - spin_lock(&nvbo->bo.bdev->fence_lock); - ttm_bo_wait(&nvbo->bo, false, false, false); - spin_unlock(&nvbo->bo.bdev->fence_lock); + if (nvbo->bo.mem.mem_type != TTM_PL_SYSTEM) nouveau_vm_unmap(vma); - } - nouveau_vm_put(vma); list_del(&vma->head); } |
