From 6dd6b7643e723b4779e59c8ad97bd5db6ff3bb12 Mon Sep 17 00:00:00 2001 From: Thomas Zimmermann Date: Mon, 18 Jan 2021 14:14:19 +0100 Subject: drm/vmwgfx: Remove reference to struct drm_device.pdev Using struct drm_device.pdev is deprecated in favor of drm_device.dev. The reference to the field was reintroduced during a rebase. Signed-off-by: Thomas Zimmermann Fixes: 9703bb329206 ("drm/vmwgfx: Switch to a managed drm device") Reviewed-by: Zack Rusin Cc: Zack Rusin Cc: Martin Krastev Cc: Roland Scheidegger Cc: VMware Graphics Cc: dri-devel@lists.freedesktop.org Link: https://patchwork.freedesktop.org/patch/msgid/20210118131420.15874-6-tzimmermann@suse.de --- drivers/gpu/drm/vmwgfx/vmwgfx_drv.c | 1 - 1 file changed, 1 deletion(-) (limited to 'drivers/gpu/drm/vmwgfx/vmwgfx_drv.c') diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c index 408847a68cf3..bc527ae2e35c 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c @@ -1516,7 +1516,6 @@ static int vmw_probe(struct pci_dev *pdev, const struct pci_device_id *ent) if (IS_ERR(vmw)) return PTR_ERR(vmw); - vmw->drm.pdev = pdev; pci_set_drvdata(pdev, &vmw->drm); ret = vmw_driver_load(vmw, ent->device); -- cgit v1.2.3 From e68cefd105c1d76e7d5a1aed4d16ad60b68214e8 Mon Sep 17 00:00:00 2001 From: Lee Jones Date: Fri, 15 Jan 2021 18:12:41 +0000 Subject: drm/vmwgfx/vmwgfx_drv: Fix some kernel-doc misdemeanours Fixes the following W=1 kernel build warning(s): drivers/gpu/drm/vmwgfx/vmwgfx_drv.c:164: warning: Function parameter or member 'ioctl' not described in 'VMW_IOCTL_DEF' drivers/gpu/drm/vmwgfx/vmwgfx_drv.c:164: warning: Function parameter or member 'func' not described in 'VMW_IOCTL_DEF' drivers/gpu/drm/vmwgfx/vmwgfx_drv.c:164: warning: Function parameter or member 'flags' not described in 'VMW_IOCTL_DEF' drivers/gpu/drm/vmwgfx/vmwgfx_drv.c:170: warning: cannot understand function prototype: 'const struct drm_ioctl_desc vmw_ioctls[] = ' drivers/gpu/drm/vmwgfx/vmwgfx_drv.c:542: warning: Function parameter or member 'dev_priv' not described in 'vmw_get_initial_size' drivers/gpu/drm/vmwgfx/vmwgfx_drv.c:611: warning: Function parameter or member 'dev_priv' not described in 'vmw_dma_masks' drivers/gpu/drm/vmwgfx/vmwgfx_drv.c:611: warning: Excess function parameter 'dev' description in 'vmw_dma_masks' Cc: VMware Graphics Cc: Roland Scheidegger Cc: David Airlie Cc: Daniel Vetter Cc: dri-devel@lists.freedesktop.org Signed-off-by: Lee Jones Signed-off-by: Zack Rusin Link: https://patchwork.freedesktop.org/patch/msgid/20210115181313.3431493-9-lee.jones@linaro.org --- drivers/gpu/drm/vmwgfx/vmwgfx_drv.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'drivers/gpu/drm/vmwgfx/vmwgfx_drv.c') diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c index bc527ae2e35c..eb997f4678de 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c @@ -153,7 +153,7 @@ DRM_IOWR(DRM_COMMAND_BASE + DRM_VMW_MSG, \ struct drm_vmw_msg_arg) -/** +/* * The core DRM version of this macro doesn't account for * DRM_COMMAND_BASE. */ @@ -161,7 +161,7 @@ #define VMW_IOCTL_DEF(ioctl, func, flags) \ [DRM_IOCTL_NR(DRM_IOCTL_##ioctl) - DRM_COMMAND_BASE] = {DRM_IOCTL_##ioctl, flags, func} -/** +/* * Ioctl definitions. */ @@ -526,7 +526,7 @@ static void vmw_release_device_late(struct vmw_private *dev_priv) vmw_fifo_release(dev_priv, &dev_priv->fifo); } -/** +/* * Sets the initial_[width|height] fields on the given vmw_private. * * It does so by reading SVGA_REG_[WIDTH|HEIGHT] regs and then @@ -599,7 +599,7 @@ static int vmw_dma_select_mode(struct vmw_private *dev_priv) /** * vmw_dma_masks - set required page- and dma masks * - * @dev: Pointer to struct drm-device + * @dev_priv: Pointer to struct drm-device * * With 32-bit we can only handle 32 bit PFNs. Optionally set that * restriction also for 64-bit systems. -- cgit v1.2.3 From 8af8a109b34fa88b8b91f25d11485b37d37549c3 Mon Sep 17 00:00:00 2001 From: Christian König Date: Thu, 1 Oct 2020 14:51:40 +0200 Subject: drm/ttm: device naming cleanup MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Rename ttm_bo_device to ttm_device. Rename ttm_bo_driver to ttm_device_funcs. Rename ttm_bo_global to ttm_global. Move global and device related functions to ttm_device.[ch]. No functional change. Signed-off-by: Christian König Acked-by: Daniel Vetter Link: https://patchwork.freedesktop.org/patch/415222/ --- drivers/gpu/drm/amd/amdgpu/amdgpu.h | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_fence.c | 6 +- drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c | 26 +- drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 8 +- drivers/gpu/drm/drm_gem_vram_helper.c | 14 +- drivers/gpu/drm/nouveau/nouveau_bo.c | 20 +- drivers/gpu/drm/nouveau/nouveau_bo.h | 2 +- drivers/gpu/drm/nouveau/nouveau_drv.h | 2 +- drivers/gpu/drm/nouveau/nouveau_sgdma.c | 6 +- drivers/gpu/drm/nouveau/nouveau_ttm.c | 10 +- drivers/gpu/drm/nouveau/nouveau_ttm.h | 8 +- drivers/gpu/drm/qxl/qxl_drv.h | 4 +- drivers/gpu/drm/qxl/qxl_release.c | 6 +- drivers/gpu/drm/qxl/qxl_ttm.c | 19 +- drivers/gpu/drm/radeon/radeon.h | 6 +- drivers/gpu/drm/radeon/radeon_object.c | 2 +- drivers/gpu/drm/radeon/radeon_ttm.c | 38 ++- drivers/gpu/drm/ttm/Makefile | 2 +- drivers/gpu/drm/ttm/ttm_agp_backend.c | 2 +- drivers/gpu/drm/ttm/ttm_bo.c | 256 ++++-------------- drivers/gpu/drm/ttm/ttm_bo_util.c | 24 +- drivers/gpu/drm/ttm/ttm_bo_vm.c | 24 +- drivers/gpu/drm/ttm/ttm_device.c | 195 ++++++++++++++ drivers/gpu/drm/ttm/ttm_execbuf_util.c | 8 +- drivers/gpu/drm/ttm/ttm_range_manager.c | 4 +- drivers/gpu/drm/ttm/ttm_resource.c | 4 +- drivers/gpu/drm/ttm/ttm_tt.c | 26 +- drivers/gpu/drm/vmwgfx/vmwgfx_blit.c | 4 +- drivers/gpu/drm/vmwgfx/vmwgfx_bo.c | 4 +- drivers/gpu/drm/vmwgfx/vmwgfx_drv.c | 16 +- drivers/gpu/drm/vmwgfx/vmwgfx_drv.h | 4 +- drivers/gpu/drm/vmwgfx/vmwgfx_resource.c | 2 +- drivers/gpu/drm/vmwgfx/vmwgfx_ttm_buffer.c | 14 +- include/drm/drm_gem_vram_helper.h | 6 +- include/drm/ttm/ttm_bo_api.h | 35 +-- include/drm/ttm/ttm_bo_driver.h | 328 +---------------------- include/drm/ttm/ttm_device.h | 319 ++++++++++++++++++++++ include/drm/ttm/ttm_resource.h | 4 +- include/drm/ttm/ttm_tt.h | 10 +- 41 files changed, 759 insertions(+), 715 deletions(-) create mode 100644 drivers/gpu/drm/ttm/ttm_device.c create mode 100644 include/drm/ttm/ttm_device.h (limited to 'drivers/gpu/drm/vmwgfx/vmwgfx_drv.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index f77443cd9c17..ab4ac3b2651e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -1053,7 +1053,7 @@ static inline struct drm_device *adev_to_drm(struct amdgpu_device *adev) return &adev->ddev; } -static inline struct amdgpu_device *amdgpu_ttm_adev(struct ttm_bo_device *bdev) +static inline struct amdgpu_device *amdgpu_ttm_adev(struct ttm_device *bdev) { return container_of(bdev, struct amdgpu_device, mman.bdev); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_fence.c index 3107b9575929..5af464933976 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_fence.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_fence.c @@ -40,13 +40,13 @@ static atomic_t fence_seq = ATOMIC_INIT(0); * All the BOs in a process share an eviction fence. When process X wants * to map VRAM memory but TTM can't find enough space, TTM will attempt to * evict BOs from its LRU list. TTM checks if the BO is valuable to evict - * by calling ttm_bo_driver->eviction_valuable(). + * by calling ttm_device_funcs->eviction_valuable(). * - * ttm_bo_driver->eviction_valuable() - will return false if the BO belongs + * ttm_device_funcs->eviction_valuable() - will return false if the BO belongs * to process X. Otherwise, it will return true to indicate BO can be * evicted by TTM. * - * If ttm_bo_driver->eviction_valuable returns true, then TTM will continue + * If ttm_device_funcs->eviction_valuable returns true, then TTM will continue * the evcition process for that BO by calling ttm_bo_evict --> amdgpu_bo_move * --> amdgpu_copy_buffer(). This sets up job in GPU scheduler. * diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c index 0db933026722..fde2d899b2c4 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c @@ -71,7 +71,7 @@ */ static int amdgpu_gart_dummy_page_init(struct amdgpu_device *adev) { - struct page *dummy_page = ttm_bo_glob.dummy_read_page; + struct page *dummy_page = ttm_glob.dummy_read_page; if (adev->dummy_page_addr) return 0; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c index 9fd2157b133a..29cfb0809634 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c @@ -61,10 +61,10 @@ #define AMDGPU_TTM_VRAM_MAX_DW_READ (size_t)128 -static int amdgpu_ttm_backend_bind(struct ttm_bo_device *bdev, +static int amdgpu_ttm_backend_bind(struct ttm_device *bdev, struct ttm_tt *ttm, struct ttm_resource *bo_mem); -static void amdgpu_ttm_backend_unbind(struct ttm_bo_device *bdev, +static void amdgpu_ttm_backend_unbind(struct ttm_device *bdev, struct ttm_tt *ttm); static int amdgpu_ttm_init_on_chip(struct amdgpu_device *adev, @@ -646,7 +646,7 @@ out: * * Called by ttm_mem_io_reserve() ultimately via ttm_bo_vm_fault() */ -static int amdgpu_ttm_io_mem_reserve(struct ttm_bo_device *bdev, struct ttm_resource *mem) +static int amdgpu_ttm_io_mem_reserve(struct ttm_device *bdev, struct ttm_resource *mem) { struct amdgpu_device *adev = amdgpu_ttm_adev(bdev); struct drm_mm_node *mm_node = mem->mm_node; @@ -893,7 +893,7 @@ void amdgpu_ttm_tt_set_user_pages(struct ttm_tt *ttm, struct page **pages) * * Called by amdgpu_ttm_backend_bind() **/ -static int amdgpu_ttm_tt_pin_userptr(struct ttm_bo_device *bdev, +static int amdgpu_ttm_tt_pin_userptr(struct ttm_device *bdev, struct ttm_tt *ttm) { struct amdgpu_device *adev = amdgpu_ttm_adev(bdev); @@ -931,7 +931,7 @@ release_sg: /* * amdgpu_ttm_tt_unpin_userptr - Unpin and unmap userptr pages */ -static void amdgpu_ttm_tt_unpin_userptr(struct ttm_bo_device *bdev, +static void amdgpu_ttm_tt_unpin_userptr(struct ttm_device *bdev, struct ttm_tt *ttm) { struct amdgpu_device *adev = amdgpu_ttm_adev(bdev); @@ -1015,7 +1015,7 @@ gart_bind_fail: * Called by ttm_tt_bind() on behalf of ttm_bo_handle_move_mem(). * This handles binding GTT memory to the device address space. */ -static int amdgpu_ttm_backend_bind(struct ttm_bo_device *bdev, +static int amdgpu_ttm_backend_bind(struct ttm_device *bdev, struct ttm_tt *ttm, struct ttm_resource *bo_mem) { @@ -1155,7 +1155,7 @@ int amdgpu_ttm_recover_gart(struct ttm_buffer_object *tbo) * Called by ttm_tt_unbind() on behalf of ttm_bo_move_ttm() and * ttm_tt_destroy(). */ -static void amdgpu_ttm_backend_unbind(struct ttm_bo_device *bdev, +static void amdgpu_ttm_backend_unbind(struct ttm_device *bdev, struct ttm_tt *ttm) { struct amdgpu_device *adev = amdgpu_ttm_adev(bdev); @@ -1180,7 +1180,7 @@ static void amdgpu_ttm_backend_unbind(struct ttm_bo_device *bdev, gtt->bound = false; } -static void amdgpu_ttm_backend_destroy(struct ttm_bo_device *bdev, +static void amdgpu_ttm_backend_destroy(struct ttm_device *bdev, struct ttm_tt *ttm) { struct amdgpu_ttm_tt *gtt = (void *)ttm; @@ -1234,7 +1234,7 @@ static struct ttm_tt *amdgpu_ttm_tt_create(struct ttm_buffer_object *bo, * Map the pages of a ttm_tt object to an address space visible * to the underlying device. */ -static int amdgpu_ttm_tt_populate(struct ttm_bo_device *bdev, +static int amdgpu_ttm_tt_populate(struct ttm_device *bdev, struct ttm_tt *ttm, struct ttm_operation_ctx *ctx) { @@ -1278,7 +1278,7 @@ static int amdgpu_ttm_tt_populate(struct ttm_bo_device *bdev, * Unmaps pages of a ttm_tt object from the device address space and * unpopulates the page array backing it. */ -static void amdgpu_ttm_tt_unpopulate(struct ttm_bo_device *bdev, +static void amdgpu_ttm_tt_unpopulate(struct ttm_device *bdev, struct ttm_tt *ttm) { struct amdgpu_ttm_tt *gtt = (void *)ttm; @@ -1603,7 +1603,7 @@ amdgpu_bo_delete_mem_notify(struct ttm_buffer_object *bo) amdgpu_bo_move_notify(bo, false, NULL); } -static struct ttm_bo_driver amdgpu_bo_driver = { +static struct ttm_device_funcs amdgpu_bo_driver = { .ttm_tt_create = &amdgpu_ttm_tt_create, .ttm_tt_populate = &amdgpu_ttm_tt_populate, .ttm_tt_unpopulate = &amdgpu_ttm_tt_unpopulate, @@ -1785,7 +1785,7 @@ int amdgpu_ttm_init(struct amdgpu_device *adev) mutex_init(&adev->mman.gtt_window_lock); /* No others user of address space so set it to 0 */ - r = ttm_bo_device_init(&adev->mman.bdev, &amdgpu_bo_driver, adev->dev, + r = ttm_device_init(&adev->mman.bdev, &amdgpu_bo_driver, adev->dev, adev_to_drm(adev)->anon_inode->i_mapping, adev_to_drm(adev)->vma_offset_manager, adev->need_swiotlb, @@ -1926,7 +1926,7 @@ void amdgpu_ttm_fini(struct amdgpu_device *adev) ttm_range_man_fini(&adev->mman.bdev, AMDGPU_PL_GDS); ttm_range_man_fini(&adev->mman.bdev, AMDGPU_PL_GWS); ttm_range_man_fini(&adev->mman.bdev, AMDGPU_PL_OA); - ttm_bo_device_release(&adev->mman.bdev); + ttm_device_fini(&adev->mman.bdev); adev->mman.initialized = false; DRM_INFO("amdgpu: ttm finalized\n"); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h index d2987536d7cd..7189f8370108 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h @@ -60,7 +60,7 @@ struct amdgpu_gtt_mgr { }; struct amdgpu_mman { - struct ttm_bo_device bdev; + struct ttm_device bdev; bool initialized; void __iomem *aper_base_kaddr; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index ad91c0c3c423..9d19078246c8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -638,15 +638,15 @@ void amdgpu_vm_move_to_lru_tail(struct amdgpu_device *adev, struct amdgpu_vm_bo_base *bo_base; if (vm->bulk_moveable) { - spin_lock(&ttm_bo_glob.lru_lock); + spin_lock(&ttm_glob.lru_lock); ttm_bo_bulk_move_lru_tail(&vm->lru_bulk_move); - spin_unlock(&ttm_bo_glob.lru_lock); + spin_unlock(&ttm_glob.lru_lock); return; } memset(&vm->lru_bulk_move, 0, sizeof(vm->lru_bulk_move)); - spin_lock(&ttm_bo_glob.lru_lock); + spin_lock(&ttm_glob.lru_lock); list_for_each_entry(bo_base, &vm->idle, vm_status) { struct amdgpu_bo *bo = bo_base->bo; @@ -660,7 +660,7 @@ void amdgpu_vm_move_to_lru_tail(struct amdgpu_device *adev, &bo->shadow->tbo.mem, &vm->lru_bulk_move); } - spin_unlock(&ttm_bo_glob.lru_lock); + spin_unlock(&ttm_glob.lru_lock); vm->bulk_moveable = true; } diff --git a/drivers/gpu/drm/drm_gem_vram_helper.c b/drivers/gpu/drm/drm_gem_vram_helper.c index 02ca22e90290..0b13c8507688 100644 --- a/drivers/gpu/drm/drm_gem_vram_helper.c +++ b/drivers/gpu/drm/drm_gem_vram_helper.c @@ -187,7 +187,7 @@ struct drm_gem_vram_object *drm_gem_vram_create(struct drm_device *dev, struct drm_gem_vram_object *gbo; struct drm_gem_object *gem; struct drm_vram_mm *vmm = dev->vram_mm; - struct ttm_bo_device *bdev; + struct ttm_device *bdev; int ret; size_t acc_size; @@ -551,7 +551,7 @@ err_drm_gem_object_put: EXPORT_SYMBOL(drm_gem_vram_fill_create_dumb); /* - * Helpers for struct ttm_bo_driver + * Helpers for struct ttm_device_funcs */ static bool drm_is_gem_vram(struct ttm_buffer_object *bo) @@ -893,7 +893,7 @@ static const struct drm_gem_object_funcs drm_gem_vram_object_funcs = { * TTM TT */ -static void bo_driver_ttm_tt_destroy(struct ttm_bo_device *bdev, struct ttm_tt *tt) +static void bo_driver_ttm_tt_destroy(struct ttm_device *bdev, struct ttm_tt *tt) { ttm_tt_destroy_common(bdev, tt); ttm_tt_fini(tt); @@ -965,7 +965,7 @@ static int bo_driver_move(struct ttm_buffer_object *bo, return drm_gem_vram_bo_driver_move(gbo, evict, ctx, new_mem); } -static int bo_driver_io_mem_reserve(struct ttm_bo_device *bdev, +static int bo_driver_io_mem_reserve(struct ttm_device *bdev, struct ttm_resource *mem) { struct drm_vram_mm *vmm = drm_vram_mm_of_bdev(bdev); @@ -985,7 +985,7 @@ static int bo_driver_io_mem_reserve(struct ttm_bo_device *bdev, return 0; } -static struct ttm_bo_driver bo_driver = { +static struct ttm_device_funcs bo_driver = { .ttm_tt_create = bo_driver_ttm_tt_create, .ttm_tt_destroy = bo_driver_ttm_tt_destroy, .eviction_valuable = ttm_bo_eviction_valuable, @@ -1036,7 +1036,7 @@ static int drm_vram_mm_init(struct drm_vram_mm *vmm, struct drm_device *dev, vmm->vram_base = vram_base; vmm->vram_size = vram_size; - ret = ttm_bo_device_init(&vmm->bdev, &bo_driver, dev->dev, + ret = ttm_device_init(&vmm->bdev, &bo_driver, dev->dev, dev->anon_inode->i_mapping, dev->vma_offset_manager, false, true); @@ -1054,7 +1054,7 @@ static int drm_vram_mm_init(struct drm_vram_mm *vmm, struct drm_device *dev, static void drm_vram_mm_cleanup(struct drm_vram_mm *vmm) { ttm_range_man_fini(&vmm->bdev, TTM_PL_VRAM); - ttm_bo_device_release(&vmm->bdev); + ttm_device_fini(&vmm->bdev); } /* diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.c b/drivers/gpu/drm/nouveau/nouveau_bo.c index 33dc886d1d6d..c177940d6e2c 100644 --- a/drivers/gpu/drm/nouveau/nouveau_bo.c +++ b/drivers/gpu/drm/nouveau/nouveau_bo.c @@ -43,9 +43,9 @@ #include #include -static int nouveau_ttm_tt_bind(struct ttm_bo_device *bdev, struct ttm_tt *ttm, +static int nouveau_ttm_tt_bind(struct ttm_device *bdev, struct ttm_tt *ttm, struct ttm_resource *reg); -static void nouveau_ttm_tt_unbind(struct ttm_bo_device *bdev, struct ttm_tt *ttm); +static void nouveau_ttm_tt_unbind(struct ttm_device *bdev, struct ttm_tt *ttm); /* * NV10-NV40 tiling helpers @@ -674,7 +674,7 @@ nouveau_ttm_tt_create(struct ttm_buffer_object *bo, uint32_t page_flags) } static int -nouveau_ttm_tt_bind(struct ttm_bo_device *bdev, struct ttm_tt *ttm, +nouveau_ttm_tt_bind(struct ttm_device *bdev, struct ttm_tt *ttm, struct ttm_resource *reg) { #if IS_ENABLED(CONFIG_AGP) @@ -690,7 +690,7 @@ nouveau_ttm_tt_bind(struct ttm_bo_device *bdev, struct ttm_tt *ttm, } static void -nouveau_ttm_tt_unbind(struct ttm_bo_device *bdev, struct ttm_tt *ttm) +nouveau_ttm_tt_unbind(struct ttm_device *bdev, struct ttm_tt *ttm) { #if IS_ENABLED(CONFIG_AGP) struct nouveau_drm *drm = nouveau_bdev(bdev); @@ -1055,7 +1055,7 @@ nouveau_ttm_io_mem_free_locked(struct nouveau_drm *drm, } static int -nouveau_ttm_io_mem_reserve(struct ttm_bo_device *bdev, struct ttm_resource *reg) +nouveau_ttm_io_mem_reserve(struct ttm_device *bdev, struct ttm_resource *reg) { struct nouveau_drm *drm = nouveau_bdev(bdev); struct nvkm_device *device = nvxx_device(&drm->client.device); @@ -1163,7 +1163,7 @@ out: } static void -nouveau_ttm_io_mem_free(struct ttm_bo_device *bdev, struct ttm_resource *reg) +nouveau_ttm_io_mem_free(struct ttm_device *bdev, struct ttm_resource *reg) { struct nouveau_drm *drm = nouveau_bdev(bdev); @@ -1223,7 +1223,7 @@ vm_fault_t nouveau_ttm_fault_reserve_notify(struct ttm_buffer_object *bo) } static int -nouveau_ttm_tt_populate(struct ttm_bo_device *bdev, +nouveau_ttm_tt_populate(struct ttm_device *bdev, struct ttm_tt *ttm, struct ttm_operation_ctx *ctx) { struct ttm_tt *ttm_dma = (void *)ttm; @@ -1247,7 +1247,7 @@ nouveau_ttm_tt_populate(struct ttm_bo_device *bdev, } static void -nouveau_ttm_tt_unpopulate(struct ttm_bo_device *bdev, +nouveau_ttm_tt_unpopulate(struct ttm_device *bdev, struct ttm_tt *ttm) { struct nouveau_drm *drm; @@ -1264,7 +1264,7 @@ nouveau_ttm_tt_unpopulate(struct ttm_bo_device *bdev, } static void -nouveau_ttm_tt_destroy(struct ttm_bo_device *bdev, +nouveau_ttm_tt_destroy(struct ttm_device *bdev, struct ttm_tt *ttm) { #if IS_ENABLED(CONFIG_AGP) @@ -1296,7 +1296,7 @@ nouveau_bo_delete_mem_notify(struct ttm_buffer_object *bo) nouveau_bo_move_ntfy(bo, false, NULL); } -struct ttm_bo_driver nouveau_bo_driver = { +struct ttm_device_funcs nouveau_bo_driver = { .ttm_tt_create = &nouveau_ttm_tt_create, .ttm_tt_populate = &nouveau_ttm_tt_populate, .ttm_tt_unpopulate = &nouveau_ttm_tt_unpopulate, diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.h b/drivers/gpu/drm/nouveau/nouveau_bo.h index 6045b85a762a..c2d3f9c48eba 100644 --- a/drivers/gpu/drm/nouveau/nouveau_bo.h +++ b/drivers/gpu/drm/nouveau/nouveau_bo.h @@ -68,7 +68,7 @@ nouveau_bo_ref(struct nouveau_bo *ref, struct nouveau_bo **pnvbo) return 0; } -extern struct ttm_bo_driver nouveau_bo_driver; +extern struct ttm_device_funcs nouveau_bo_driver; void nouveau_bo_move_init(struct nouveau_drm *); struct nouveau_bo *nouveau_bo_alloc(struct nouveau_cli *, u64 *size, int *align, diff --git a/drivers/gpu/drm/nouveau/nouveau_drv.h b/drivers/gpu/drm/nouveau/nouveau_drv.h index c802d3d1ba39..edf9d1ee9d58 100644 --- a/drivers/gpu/drm/nouveau/nouveau_drv.h +++ b/drivers/gpu/drm/nouveau/nouveau_drv.h @@ -151,7 +151,7 @@ struct nouveau_drm { /* TTM interface support */ struct { - struct ttm_bo_device bdev; + struct ttm_device bdev; atomic_t validate_sequence; int (*move)(struct nouveau_channel *, struct ttm_buffer_object *, diff --git a/drivers/gpu/drm/nouveau/nouveau_sgdma.c b/drivers/gpu/drm/nouveau/nouveau_sgdma.c index 1cf52635ea74..256ec5b35473 100644 --- a/drivers/gpu/drm/nouveau/nouveau_sgdma.c +++ b/drivers/gpu/drm/nouveau/nouveau_sgdma.c @@ -16,7 +16,7 @@ struct nouveau_sgdma_be { }; void -nouveau_sgdma_destroy(struct ttm_bo_device *bdev, struct ttm_tt *ttm) +nouveau_sgdma_destroy(struct ttm_device *bdev, struct ttm_tt *ttm) { struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)ttm; @@ -29,7 +29,7 @@ nouveau_sgdma_destroy(struct ttm_bo_device *bdev, struct ttm_tt *ttm) } int -nouveau_sgdma_bind(struct ttm_bo_device *bdev, struct ttm_tt *ttm, struct ttm_resource *reg) +nouveau_sgdma_bind(struct ttm_device *bdev, struct ttm_tt *ttm, struct ttm_resource *reg) { struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)ttm; struct nouveau_drm *drm = nouveau_bdev(bdev); @@ -56,7 +56,7 @@ nouveau_sgdma_bind(struct ttm_bo_device *bdev, struct ttm_tt *ttm, struct ttm_re } void -nouveau_sgdma_unbind(struct ttm_bo_device *bdev, struct ttm_tt *ttm) +nouveau_sgdma_unbind(struct ttm_device *bdev, struct ttm_tt *ttm) { struct nouveau_sgdma_be *nvbe = (struct nouveau_sgdma_be *)ttm; if (nvbe->mem) { diff --git a/drivers/gpu/drm/nouveau/nouveau_ttm.c b/drivers/gpu/drm/nouveau/nouveau_ttm.c index a37bc3d7b38b..495288182c2d 100644 --- a/drivers/gpu/drm/nouveau/nouveau_ttm.c +++ b/drivers/gpu/drm/nouveau/nouveau_ttm.c @@ -324,10 +324,10 @@ nouveau_ttm_init(struct nouveau_drm *drm) need_swiotlb = !!swiotlb_nr_tbl(); #endif - ret = ttm_bo_device_init(&drm->ttm.bdev, &nouveau_bo_driver, - drm->dev->dev, dev->anon_inode->i_mapping, - dev->vma_offset_manager, need_swiotlb, - drm->client.mmu.dmabits <= 32); + ret = ttm_device_init(&drm->ttm.bdev, &nouveau_bo_driver, drm->dev->dev, + dev->anon_inode->i_mapping, + dev->vma_offset_manager, need_swiotlb, + drm->client.mmu.dmabits <= 32); if (ret) { NV_ERROR(drm, "error initialising bo driver, %d\n", ret); return ret; @@ -377,7 +377,7 @@ nouveau_ttm_fini(struct nouveau_drm *drm) nouveau_ttm_fini_vram(drm); nouveau_ttm_fini_gtt(drm); - ttm_bo_device_release(&drm->ttm.bdev); + ttm_device_fini(&drm->ttm.bdev); arch_phys_wc_del(drm->ttm.mtrr); drm->ttm.mtrr = 0; diff --git a/drivers/gpu/drm/nouveau/nouveau_ttm.h b/drivers/gpu/drm/nouveau/nouveau_ttm.h index 69552049bb96..dbf6dc238efd 100644 --- a/drivers/gpu/drm/nouveau/nouveau_ttm.h +++ b/drivers/gpu/drm/nouveau/nouveau_ttm.h @@ -3,7 +3,7 @@ #define __NOUVEAU_TTM_H__ static inline struct nouveau_drm * -nouveau_bdev(struct ttm_bo_device *bd) +nouveau_bdev(struct ttm_device *bd) { return container_of(bd, struct nouveau_drm, ttm.bdev); } @@ -22,7 +22,7 @@ int nouveau_ttm_mmap(struct file *, struct vm_area_struct *); int nouveau_ttm_global_init(struct nouveau_drm *); void nouveau_ttm_global_release(struct nouveau_drm *); -int nouveau_sgdma_bind(struct ttm_bo_device *bdev, struct ttm_tt *ttm, struct ttm_resource *reg); -void nouveau_sgdma_unbind(struct ttm_bo_device *bdev, struct ttm_tt *ttm); -void nouveau_sgdma_destroy(struct ttm_bo_device *bdev, struct ttm_tt *ttm); +int nouveau_sgdma_bind(struct ttm_device *bdev, struct ttm_tt *ttm, struct ttm_resource *reg); +void nouveau_sgdma_unbind(struct ttm_device *bdev, struct ttm_tt *ttm); +void nouveau_sgdma_destroy(struct ttm_device *bdev, struct ttm_tt *ttm); #endif diff --git a/drivers/gpu/drm/qxl/qxl_drv.h b/drivers/gpu/drm/qxl/qxl_drv.h index 83b54f0dad61..01354b43c413 100644 --- a/drivers/gpu/drm/qxl/qxl_drv.h +++ b/drivers/gpu/drm/qxl/qxl_drv.h @@ -125,7 +125,7 @@ struct qxl_output { #define drm_encoder_to_qxl_output(x) container_of(x, struct qxl_output, enc) struct qxl_mman { - struct ttm_bo_device bdev; + struct ttm_device bdev; }; struct qxl_memslot { @@ -335,7 +335,7 @@ int qxl_mode_dumb_mmap(struct drm_file *filp, /* qxl ttm */ int qxl_ttm_init(struct qxl_device *qdev); void qxl_ttm_fini(struct qxl_device *qdev); -int qxl_ttm_io_mem_reserve(struct ttm_bo_device *bdev, +int qxl_ttm_io_mem_reserve(struct ttm_device *bdev, struct ttm_resource *mem); /* qxl image */ diff --git a/drivers/gpu/drm/qxl/qxl_release.c b/drivers/gpu/drm/qxl/qxl_release.c index 0fcfc952d5e9..c52412724c26 100644 --- a/drivers/gpu/drm/qxl/qxl_release.c +++ b/drivers/gpu/drm/qxl/qxl_release.c @@ -429,7 +429,7 @@ void qxl_release_unmap(struct qxl_device *qdev, void qxl_release_fence_buffer_objects(struct qxl_release *release) { struct ttm_buffer_object *bo; - struct ttm_bo_device *bdev; + struct ttm_device *bdev; struct ttm_validate_buffer *entry; struct qxl_device *qdev; @@ -450,7 +450,7 @@ void qxl_release_fence_buffer_objects(struct qxl_release *release) release->id | 0xf0000000, release->base.seqno); trace_dma_fence_emit(&release->base); - spin_lock(&ttm_bo_glob.lru_lock); + spin_lock(&ttm_glob.lru_lock); list_for_each_entry(entry, &release->bos, head) { bo = entry->bo; @@ -459,7 +459,7 @@ void qxl_release_fence_buffer_objects(struct qxl_release *release) ttm_bo_move_to_lru_tail(bo, &bo->mem, NULL); dma_resv_unlock(bo->base.resv); } - spin_unlock(&ttm_bo_glob.lru_lock); + spin_unlock(&ttm_glob.lru_lock); ww_acquire_fini(&release->ticket); } diff --git a/drivers/gpu/drm/qxl/qxl_ttm.c b/drivers/gpu/drm/qxl/qxl_ttm.c index 33c09dc94f8b..b7f77eb685cb 100644 --- a/drivers/gpu/drm/qxl/qxl_ttm.c +++ b/drivers/gpu/drm/qxl/qxl_ttm.c @@ -36,7 +36,7 @@ #include "qxl_drv.h" #include "qxl_object.h" -static struct qxl_device *qxl_get_qdev(struct ttm_bo_device *bdev) +static struct qxl_device *qxl_get_qdev(struct ttm_device *bdev) { struct qxl_mman *mman; struct qxl_device *qdev; @@ -69,7 +69,7 @@ static void qxl_evict_flags(struct ttm_buffer_object *bo, *placement = qbo->placement; } -int qxl_ttm_io_mem_reserve(struct ttm_bo_device *bdev, +int qxl_ttm_io_mem_reserve(struct ttm_device *bdev, struct ttm_resource *mem) { struct qxl_device *qdev = qxl_get_qdev(bdev); @@ -98,8 +98,7 @@ int qxl_ttm_io_mem_reserve(struct ttm_bo_device *bdev, /* * TTM backend functions. */ -static void qxl_ttm_backend_destroy(struct ttm_bo_device *bdev, - struct ttm_tt *ttm) +static void qxl_ttm_backend_destroy(struct ttm_device *bdev, struct ttm_tt *ttm) { ttm_tt_destroy_common(bdev, ttm); ttm_tt_fini(ttm); @@ -170,7 +169,7 @@ static void qxl_bo_delete_mem_notify(struct ttm_buffer_object *bo) qxl_bo_move_notify(bo, false, NULL); } -static struct ttm_bo_driver qxl_bo_driver = { +static struct ttm_device_funcs qxl_bo_driver = { .ttm_tt_create = &qxl_ttm_tt_create, .ttm_tt_destroy = &qxl_ttm_backend_destroy, .eviction_valuable = ttm_bo_eviction_valuable, @@ -193,10 +192,10 @@ int qxl_ttm_init(struct qxl_device *qdev) int num_io_pages; /* != rom->num_io_pages, we include surface0 */ /* No others user of address space so set it to 0 */ - r = ttm_bo_device_init(&qdev->mman.bdev, &qxl_bo_driver, NULL, - qdev->ddev.anon_inode->i_mapping, - qdev->ddev.vma_offset_manager, - false, false); + r = ttm_device_init(&qdev->mman.bdev, &qxl_bo_driver, NULL, + qdev->ddev.anon_inode->i_mapping, + qdev->ddev.vma_offset_manager, + false, false); if (r) { DRM_ERROR("failed initializing buffer object driver(%d).\n", r); return r; @@ -227,7 +226,7 @@ void qxl_ttm_fini(struct qxl_device *qdev) { ttm_range_man_fini(&qdev->mman.bdev, TTM_PL_VRAM); ttm_range_man_fini(&qdev->mman.bdev, TTM_PL_PRIV); - ttm_bo_device_release(&qdev->mman.bdev); + ttm_device_fini(&qdev->mman.bdev); DRM_INFO("qxl: ttm finalized\n"); } diff --git a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h index f09989bdce98..449ba0095ed9 100644 --- a/drivers/gpu/drm/radeon/radeon.h +++ b/drivers/gpu/drm/radeon/radeon.h @@ -451,7 +451,7 @@ struct radeon_surface_reg { * TTM. */ struct radeon_mman { - struct ttm_bo_device bdev; + struct ttm_device bdev; bool initialized; #if defined(CONFIG_DEBUG_FS) @@ -2822,7 +2822,7 @@ extern int radeon_ttm_tt_set_userptr(struct radeon_device *rdev, uint32_t flags); extern bool radeon_ttm_tt_has_userptr(struct radeon_device *rdev, struct ttm_tt *ttm); extern bool radeon_ttm_tt_is_readonly(struct radeon_device *rdev, struct ttm_tt *ttm); -bool radeon_ttm_tt_is_bound(struct ttm_bo_device *bdev, struct ttm_tt *ttm); +bool radeon_ttm_tt_is_bound(struct ttm_device *bdev, struct ttm_tt *ttm); extern void radeon_vram_location(struct radeon_device *rdev, struct radeon_mc *mc, u64 base); extern void radeon_gtt_location(struct radeon_device *rdev, struct radeon_mc *mc); extern int radeon_resume_kms(struct drm_device *dev, bool resume, bool fbcon); @@ -2832,7 +2832,7 @@ extern void radeon_ttm_set_active_vram_size(struct radeon_device *rdev, u64 size extern void radeon_program_register_sequence(struct radeon_device *rdev, const u32 *registers, const u32 array_size); -struct radeon_device *radeon_get_rdev(struct ttm_bo_device *bdev); +struct radeon_device *radeon_get_rdev(struct ttm_device *bdev); /* KMS */ diff --git a/drivers/gpu/drm/radeon/radeon_object.c b/drivers/gpu/drm/radeon/radeon_object.c index 9b81786782de..6a336284466f 100644 --- a/drivers/gpu/drm/radeon/radeon_object.c +++ b/drivers/gpu/drm/radeon/radeon_object.c @@ -372,7 +372,7 @@ void radeon_bo_unpin(struct radeon_bo *bo) int radeon_bo_evict_vram(struct radeon_device *rdev) { - struct ttm_bo_device *bdev = &rdev->mman.bdev; + struct ttm_device *bdev = &rdev->mman.bdev; struct ttm_resource_manager *man; /* late 2.6.33 fix IGP hibernate - we need pm ops to do this correct */ diff --git a/drivers/gpu/drm/radeon/radeon_ttm.c b/drivers/gpu/drm/radeon/radeon_ttm.c index c6d575f50c48..5ed6a48246d6 100644 --- a/drivers/gpu/drm/radeon/radeon_ttm.c +++ b/drivers/gpu/drm/radeon/radeon_ttm.c @@ -55,13 +55,11 @@ static int radeon_ttm_debugfs_init(struct radeon_device *rdev); static void radeon_ttm_debugfs_fini(struct radeon_device *rdev); -static int radeon_ttm_tt_bind(struct ttm_bo_device *bdev, - struct ttm_tt *ttm, +static int radeon_ttm_tt_bind(struct ttm_device *bdev, struct ttm_tt *ttm, struct ttm_resource *bo_mem); -static void radeon_ttm_tt_unbind(struct ttm_bo_device *bdev, - struct ttm_tt *ttm); +static void radeon_ttm_tt_unbind(struct ttm_device *bdev, struct ttm_tt *ttm); -struct radeon_device *radeon_get_rdev(struct ttm_bo_device *bdev) +struct radeon_device *radeon_get_rdev(struct ttm_device *bdev) { struct radeon_mman *mman; struct radeon_device *rdev; @@ -280,7 +278,7 @@ out: return 0; } -static int radeon_ttm_io_mem_reserve(struct ttm_bo_device *bdev, struct ttm_resource *mem) +static int radeon_ttm_io_mem_reserve(struct ttm_device *bdev, struct ttm_resource *mem) { struct radeon_device *rdev = radeon_get_rdev(bdev); size_t bus_size = (size_t)mem->num_pages << PAGE_SHIFT; @@ -347,7 +345,7 @@ struct radeon_ttm_tt { }; /* prepare the sg table with the user pages */ -static int radeon_ttm_tt_pin_userptr(struct ttm_bo_device *bdev, struct ttm_tt *ttm) +static int radeon_ttm_tt_pin_userptr(struct ttm_device *bdev, struct ttm_tt *ttm) { struct radeon_device *rdev = radeon_get_rdev(bdev); struct radeon_ttm_tt *gtt = (void *)ttm; @@ -408,7 +406,7 @@ release_pages: return r; } -static void radeon_ttm_tt_unpin_userptr(struct ttm_bo_device *bdev, struct ttm_tt *ttm) +static void radeon_ttm_tt_unpin_userptr(struct ttm_device *bdev, struct ttm_tt *ttm) { struct radeon_device *rdev = radeon_get_rdev(bdev); struct radeon_ttm_tt *gtt = (void *)ttm; @@ -444,7 +442,7 @@ static bool radeon_ttm_backend_is_bound(struct ttm_tt *ttm) return (gtt->bound); } -static int radeon_ttm_backend_bind(struct ttm_bo_device *bdev, +static int radeon_ttm_backend_bind(struct ttm_device *bdev, struct ttm_tt *ttm, struct ttm_resource *bo_mem) { @@ -480,7 +478,7 @@ static int radeon_ttm_backend_bind(struct ttm_bo_device *bdev, return 0; } -static void radeon_ttm_backend_unbind(struct ttm_bo_device *bdev, struct ttm_tt *ttm) +static void radeon_ttm_backend_unbind(struct ttm_device *bdev, struct ttm_tt *ttm) { struct radeon_ttm_tt *gtt = (void *)ttm; struct radeon_device *rdev = radeon_get_rdev(bdev); @@ -495,7 +493,7 @@ static void radeon_ttm_backend_unbind(struct ttm_bo_device *bdev, struct ttm_tt gtt->bound = false; } -static void radeon_ttm_backend_destroy(struct ttm_bo_device *bdev, struct ttm_tt *ttm) +static void radeon_ttm_backend_destroy(struct ttm_device *bdev, struct ttm_tt *ttm) { struct radeon_ttm_tt *gtt = (void *)ttm; @@ -554,7 +552,7 @@ static struct radeon_ttm_tt *radeon_ttm_tt_to_gtt(struct radeon_device *rdev, return container_of(ttm, struct radeon_ttm_tt, ttm); } -static int radeon_ttm_tt_populate(struct ttm_bo_device *bdev, +static int radeon_ttm_tt_populate(struct ttm_device *bdev, struct ttm_tt *ttm, struct ttm_operation_ctx *ctx) { @@ -580,7 +578,7 @@ static int radeon_ttm_tt_populate(struct ttm_bo_device *bdev, return ttm_pool_alloc(&rdev->mman.bdev.pool, ttm, ctx); } -static void radeon_ttm_tt_unpopulate(struct ttm_bo_device *bdev, struct ttm_tt *ttm) +static void radeon_ttm_tt_unpopulate(struct ttm_device *bdev, struct ttm_tt *ttm) { struct radeon_device *rdev = radeon_get_rdev(bdev); struct radeon_ttm_tt *gtt = radeon_ttm_tt_to_gtt(rdev, ttm); @@ -613,7 +611,7 @@ int radeon_ttm_tt_set_userptr(struct radeon_device *rdev, return 0; } -bool radeon_ttm_tt_is_bound(struct ttm_bo_device *bdev, +bool radeon_ttm_tt_is_bound(struct ttm_device *bdev, struct ttm_tt *ttm) { #if IS_ENABLED(CONFIG_AGP) @@ -624,7 +622,7 @@ bool radeon_ttm_tt_is_bound(struct ttm_bo_device *bdev, return radeon_ttm_backend_is_bound(ttm); } -static int radeon_ttm_tt_bind(struct ttm_bo_device *bdev, +static int radeon_ttm_tt_bind(struct ttm_device *bdev, struct ttm_tt *ttm, struct ttm_resource *bo_mem) { @@ -642,7 +640,7 @@ static int radeon_ttm_tt_bind(struct ttm_bo_device *bdev, return radeon_ttm_backend_bind(bdev, ttm, bo_mem); } -static void radeon_ttm_tt_unbind(struct ttm_bo_device *bdev, +static void radeon_ttm_tt_unbind(struct ttm_device *bdev, struct ttm_tt *ttm) { #if IS_ENABLED(CONFIG_AGP) @@ -656,7 +654,7 @@ static void radeon_ttm_tt_unbind(struct ttm_bo_device *bdev, radeon_ttm_backend_unbind(bdev, ttm); } -static void radeon_ttm_tt_destroy(struct ttm_bo_device *bdev, +static void radeon_ttm_tt_destroy(struct ttm_device *bdev, struct ttm_tt *ttm) { #if IS_ENABLED(CONFIG_AGP) @@ -700,7 +698,7 @@ radeon_bo_delete_mem_notify(struct ttm_buffer_object *bo) radeon_bo_move_notify(bo, false, NULL); } -static struct ttm_bo_driver radeon_bo_driver = { +static struct ttm_device_funcs radeon_bo_driver = { .ttm_tt_create = &radeon_ttm_tt_create, .ttm_tt_populate = &radeon_ttm_tt_populate, .ttm_tt_unpopulate = &radeon_ttm_tt_unpopulate, @@ -718,7 +716,7 @@ int radeon_ttm_init(struct radeon_device *rdev) int r; /* No others user of address space so set it to 0 */ - r = ttm_bo_device_init(&rdev->mman.bdev, &radeon_bo_driver, rdev->dev, + r = ttm_device_init(&rdev->mman.bdev, &radeon_bo_driver, rdev->dev, rdev->ddev->anon_inode->i_mapping, rdev->ddev->vma_offset_manager, rdev->need_swiotlb, @@ -791,7 +789,7 @@ void radeon_ttm_fini(struct radeon_device *rdev) } ttm_range_man_fini(&rdev->mman.bdev, TTM_PL_VRAM); ttm_range_man_fini(&rdev->mman.bdev, TTM_PL_TT); - ttm_bo_device_release(&rdev->mman.bdev); + ttm_device_fini(&rdev->mman.bdev); radeon_gart_fini(rdev); rdev->mman.initialized = false; DRM_INFO("radeon: ttm finalized\n"); diff --git a/drivers/gpu/drm/ttm/Makefile b/drivers/gpu/drm/ttm/Makefile index b6f5f87b270f..8e6437eadabe 100644 --- a/drivers/gpu/drm/ttm/Makefile +++ b/drivers/gpu/drm/ttm/Makefile @@ -5,7 +5,7 @@ ttm-y := ttm_memory.o ttm_tt.o ttm_bo.o \ ttm_bo_util.o ttm_bo_vm.o ttm_module.o \ ttm_execbuf_util.o ttm_range_manager.o \ - ttm_resource.o ttm_pool.o + ttm_resource.o ttm_pool.o ttm_device.o ttm-$(CONFIG_AGP) += ttm_agp_backend.o obj-$(CONFIG_DRM_TTM) += ttm.o diff --git a/drivers/gpu/drm/ttm/ttm_agp_backend.c b/drivers/gpu/drm/ttm/ttm_agp_backend.c index 8f9fa4188897..0226ae69d3ab 100644 --- a/drivers/gpu/drm/ttm/ttm_agp_backend.c +++ b/drivers/gpu/drm/ttm/ttm_agp_backend.c @@ -49,7 +49,7 @@ struct ttm_agp_backend { int ttm_agp_bind(struct ttm_tt *ttm, struct ttm_resource *bo_mem) { struct ttm_agp_backend *agp_be = container_of(ttm, struct ttm_agp_backend, ttm); - struct page *dummy_read_page = ttm_bo_glob.dummy_read_page; + struct page *dummy_read_page = ttm_glob.dummy_read_page; struct drm_mm_node *node = bo_mem->mm_node; struct agp_memory *mem; int ret, cached = ttm->caching == ttm_cached; diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index c289a6a37ff9..20256797f3a6 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -44,14 +44,6 @@ #include "ttm_module.h" -/* - * ttm_global_mutex - protecting the global BO state - */ -DEFINE_MUTEX(ttm_global_mutex); -unsigned ttm_bo_glob_use_count; -struct ttm_bo_global ttm_bo_glob; -EXPORT_SYMBOL(ttm_bo_glob); - /* default destructor */ static void ttm_bo_default_destroy(struct ttm_buffer_object *bo) { @@ -79,13 +71,13 @@ static void ttm_bo_mem_space_debug(struct ttm_buffer_object *bo, static void ttm_bo_del_from_lru(struct ttm_buffer_object *bo) { - struct ttm_bo_device *bdev = bo->bdev; + struct ttm_device *bdev = bo->bdev; list_del_init(&bo->swap); list_del_init(&bo->lru); - if (bdev->driver->del_from_lru_notify) - bdev->driver->del_from_lru_notify(bo); + if (bdev->funcs->del_from_lru_notify) + bdev->funcs->del_from_lru_notify(bo); } static void ttm_bo_bulk_move_set_pos(struct ttm_lru_bulk_move_pos *pos, @@ -100,7 +92,7 @@ void ttm_bo_move_to_lru_tail(struct ttm_buffer_object *bo, struct ttm_resource *mem, struct ttm_lru_bulk_move *bulk) { - struct ttm_bo_device *bdev = bo->bdev; + struct ttm_device *bdev = bo->bdev; struct ttm_resource_manager *man; dma_resv_assert_held(bo->base.resv); @@ -117,12 +109,12 @@ void ttm_bo_move_to_lru_tail(struct ttm_buffer_object *bo, TTM_PAGE_FLAG_SWAPPED))) { struct list_head *swap; - swap = &ttm_bo_glob.swap_lru[bo->priority]; + swap = &ttm_glob.swap_lru[bo->priority]; list_move_tail(&bo->swap, swap); } - if (bdev->driver->del_from_lru_notify) - bdev->driver->del_from_lru_notify(bo); + if (bdev->funcs->del_from_lru_notify) + bdev->funcs->del_from_lru_notify(bo); if (bulk && !bo->pin_count) { switch (bo->mem.mem_type) { @@ -185,7 +177,7 @@ void ttm_bo_bulk_move_lru_tail(struct ttm_lru_bulk_move *bulk) dma_resv_assert_held(pos->first->base.resv); dma_resv_assert_held(pos->last->base.resv); - lru = &ttm_bo_glob.swap_lru[i]; + lru = &ttm_glob.swap_lru[i]; list_bulk_move_tail(lru, &pos->first->swap, &pos->last->swap); } } @@ -196,7 +188,7 @@ static int ttm_bo_handle_move_mem(struct ttm_buffer_object *bo, struct ttm_operation_ctx *ctx, struct ttm_place *hop) { - struct ttm_bo_device *bdev = bo->bdev; + struct ttm_device *bdev = bo->bdev; struct ttm_resource_manager *old_man = ttm_manager_type(bdev, bo->mem.mem_type); struct ttm_resource_manager *new_man = ttm_manager_type(bdev, mem->mem_type); int ret; @@ -222,7 +214,7 @@ static int ttm_bo_handle_move_mem(struct ttm_buffer_object *bo, } } - ret = bdev->driver->move(bo, evict, ctx, mem, hop); + ret = bdev->funcs->move(bo, evict, ctx, mem, hop); if (ret) { if (ret == -EMULTIHOP) return ret; @@ -250,8 +242,8 @@ out_err: static void ttm_bo_cleanup_memtype_use(struct ttm_buffer_object *bo) { - if (bo->bdev->driver->delete_mem_notify) - bo->bdev->driver->delete_mem_notify(bo); + if (bo->bdev->funcs->delete_mem_notify) + bo->bdev->funcs->delete_mem_notify(bo); ttm_bo_tt_destroy(bo); ttm_resource_free(bo, &bo->mem); @@ -276,9 +268,9 @@ static int ttm_bo_individualize_resv(struct ttm_buffer_object *bo) * reference it any more. The only tricky case is the trylock on * the resv object while holding the lru_lock. */ - spin_lock(&ttm_bo_glob.lru_lock); + spin_lock(&ttm_glob.lru_lock); bo->base.resv = &bo->base._resv; - spin_unlock(&ttm_bo_glob.lru_lock); + spin_unlock(&ttm_glob.lru_lock); } return r; @@ -337,7 +329,7 @@ static int ttm_bo_cleanup_refs(struct ttm_buffer_object *bo, if (unlock_resv) dma_resv_unlock(bo->base.resv); - spin_unlock(&ttm_bo_glob.lru_lock); + spin_unlock(&ttm_glob.lru_lock); lret = dma_resv_wait_timeout_rcu(resv, true, interruptible, 30 * HZ); @@ -347,7 +339,7 @@ static int ttm_bo_cleanup_refs(struct ttm_buffer_object *bo, else if (lret == 0) return -EBUSY; - spin_lock(&ttm_bo_glob.lru_lock); + spin_lock(&ttm_glob.lru_lock); if (unlock_resv && !dma_resv_trylock(bo->base.resv)) { /* * We raced, and lost, someone else holds the reservation now, @@ -357,7 +349,7 @@ static int ttm_bo_cleanup_refs(struct ttm_buffer_object *bo, * delayed destruction would succeed, so just return success * here. */ - spin_unlock(&ttm_bo_glob.lru_lock); + spin_unlock(&ttm_glob.lru_lock); return 0; } ret = 0; @@ -366,13 +358,13 @@ static int ttm_bo_cleanup_refs(struct ttm_buffer_object *bo, if (ret || unlikely(list_empty(&bo->ddestroy))) { if (unlock_resv) dma_resv_unlock(bo->base.resv); - spin_unlock(&ttm_bo_glob.lru_lock); + spin_unlock(&ttm_glob.lru_lock); return ret; } ttm_bo_del_from_lru(bo); list_del_init(&bo->ddestroy); - spin_unlock(&ttm_bo_glob.lru_lock); + spin_unlock(&ttm_glob.lru_lock); ttm_bo_cleanup_memtype_use(bo); if (unlock_resv) @@ -387,9 +379,9 @@ static int ttm_bo_cleanup_refs(struct ttm_buffer_object *bo, * Traverse the delayed list, and call ttm_bo_cleanup_refs on all * encountered buffers. */ -static bool ttm_bo_delayed_delete(struct ttm_bo_device *bdev, bool remove_all) +bool ttm_bo_delayed_delete(struct ttm_device *bdev, bool remove_all) { - struct ttm_bo_global *glob = &ttm_bo_glob; + struct ttm_global *glob = &ttm_glob; struct list_head removed; bool empty; @@ -428,21 +420,11 @@ static bool ttm_bo_delayed_delete(struct ttm_bo_device *bdev, bool remove_all) return empty; } -static void ttm_bo_delayed_workqueue(struct work_struct *work) -{ - struct ttm_bo_device *bdev = - container_of(work, struct ttm_bo_device, wq.work); - - if (!ttm_bo_delayed_delete(bdev, false)) - schedule_delayed_work(&bdev->wq, - ((HZ / 100) < 1) ? 1 : HZ / 100); -} - static void ttm_bo_release(struct kref *kref) { struct ttm_buffer_object *bo = container_of(kref, struct ttm_buffer_object, kref); - struct ttm_bo_device *bdev = bo->bdev; + struct ttm_device *bdev = bo->bdev; size_t acc_size = bo->acc_size; int ret; @@ -456,8 +438,8 @@ static void ttm_bo_release(struct kref *kref) 30 * HZ); } - if (bo->bdev->driver->release_notify) - bo->bdev->driver->release_notify(bo); + if (bo->bdev->funcs->release_notify) + bo->bdev->funcs->release_notify(bo); drm_vma_offset_remove(bdev->vma_manager, &bo->base.vma_node); ttm_mem_io_free(bdev, &bo->mem); @@ -469,7 +451,7 @@ static void ttm_bo_release(struct kref *kref) ttm_bo_flush_all_fences(bo); bo->deleted = true; - spin_lock(&ttm_bo_glob.lru_lock); + spin_lock(&ttm_glob.lru_lock); /* * Make pinned bos immediately available to @@ -483,22 +465,22 @@ static void ttm_bo_release(struct kref *kref) kref_init(&bo->kref); list_add_tail(&bo->ddestroy, &bdev->ddestroy); - spin_unlock(&ttm_bo_glob.lru_lock); + spin_unlock(&ttm_glob.lru_lock); schedule_delayed_work(&bdev->wq, ((HZ / 100) < 1) ? 1 : HZ / 100); return; } - spin_lock(&ttm_bo_glob.lru_lock); + spin_lock(&ttm_glob.lru_lock); ttm_bo_del_from_lru(bo); list_del(&bo->ddestroy); - spin_unlock(&ttm_bo_glob.lru_lock); + spin_unlock(&ttm_glob.lru_lock); ttm_bo_cleanup_memtype_use(bo); dma_resv_unlock(bo->base.resv); - atomic_dec(&ttm_bo_glob.bo_count); + atomic_dec(&ttm_glob.bo_count); dma_fence_put(bo->moving); if (!ttm_bo_uses_embedded_gem_object(bo)) dma_resv_fini(&bo->base._resv); @@ -512,13 +494,13 @@ void ttm_bo_put(struct ttm_buffer_object *bo) } EXPORT_SYMBOL(ttm_bo_put); -int ttm_bo_lock_delayed_workqueue(struct ttm_bo_device *bdev) +int ttm_bo_lock_delayed_workqueue(struct ttm_device *bdev) { return cancel_delayed_work_sync(&bdev->wq); } EXPORT_SYMBOL(ttm_bo_lock_delayed_workqueue); -void ttm_bo_unlock_delayed_workqueue(struct ttm_bo_device *bdev, int resched) +void ttm_bo_unlock_delayed_workqueue(struct ttm_device *bdev, int resched) { if (resched) schedule_delayed_work(&bdev->wq, @@ -529,7 +511,7 @@ EXPORT_SYMBOL(ttm_bo_unlock_delayed_workqueue); static int ttm_bo_evict(struct ttm_buffer_object *bo, struct ttm_operation_ctx *ctx) { - struct ttm_bo_device *bdev = bo->bdev; + struct ttm_device *bdev = bo->bdev; struct ttm_resource evict_mem; struct ttm_placement placement; struct ttm_place hop; @@ -541,7 +523,7 @@ static int ttm_bo_evict(struct ttm_buffer_object *bo, placement.num_placement = 0; placement.num_busy_placement = 0; - bdev->driver->evict_flags(bo, &placement); + bdev->funcs->evict_flags(bo, &placement); if (!placement.num_placement && !placement.num_busy_placement) { ttm_bo_wait(bo, false, false); @@ -657,7 +639,7 @@ static int ttm_mem_evict_wait_busy(struct ttm_buffer_object *busy_bo, return r == -EDEADLK ? -EBUSY : r; } -int ttm_mem_evict_first(struct ttm_bo_device *bdev, +int ttm_mem_evict_first(struct ttm_device *bdev, struct ttm_resource_manager *man, const struct ttm_place *place, struct ttm_operation_ctx *ctx, @@ -668,7 +650,7 @@ int ttm_mem_evict_first(struct ttm_bo_device *bdev, unsigned i; int ret; - spin_lock(&ttm_bo_glob.lru_lock); + spin_lock(&ttm_glob.lru_lock); for (i = 0; i < TTM_MAX_BO_PRIORITY; ++i) { list_for_each_entry(bo, &man->lru[i], lru) { bool busy; @@ -681,7 +663,7 @@ int ttm_mem_evict_first(struct ttm_bo_device *bdev, continue; } - if (place && !bdev->driver->eviction_valuable(bo, + if (place && !bdev->funcs->eviction_valuable(bo, place)) { if (locked) dma_resv_unlock(bo->base.resv); @@ -705,7 +687,7 @@ int ttm_mem_evict_first(struct ttm_bo_device *bdev, if (!bo) { if (busy_bo && !ttm_bo_get_unless_zero(busy_bo)) busy_bo = NULL; - spin_unlock(&ttm_bo_glob.lru_lock); + spin_unlock(&ttm_glob.lru_lock); ret = ttm_mem_evict_wait_busy(busy_bo, ctx, ticket); if (busy_bo) ttm_bo_put(busy_bo); @@ -719,7 +701,7 @@ int ttm_mem_evict_first(struct ttm_bo_device *bdev, return ret; } - spin_unlock(&ttm_bo_glob.lru_lock); + spin_unlock(&ttm_glob.lru_lock); ret = ttm_bo_evict(bo, ctx); if (locked) @@ -774,7 +756,7 @@ static int ttm_bo_mem_force_space(struct ttm_buffer_object *bo, struct ttm_resource *mem, struct ttm_operation_ctx *ctx) { - struct ttm_bo_device *bdev = bo->bdev; + struct ttm_device *bdev = bo->bdev; struct ttm_resource_manager *man = ttm_manager_type(bdev, mem->mem_type); struct ww_acquire_ctx *ticket; int ret; @@ -809,7 +791,7 @@ static int ttm_bo_mem_placement(struct ttm_buffer_object *bo, const struct ttm_place *place, struct ttm_resource *mem) { - struct ttm_bo_device *bdev = bo->bdev; + struct ttm_device *bdev = bo->bdev; struct ttm_resource_manager *man; man = ttm_manager_type(bdev, place->mem_type); @@ -819,9 +801,9 @@ static int ttm_bo_mem_placement(struct ttm_buffer_object *bo, mem->mem_type = place->mem_type; mem->placement = place->flags; - spin_lock(&ttm_bo_glob.lru_lock); + spin_lock(&ttm_glob.lru_lock); ttm_bo_move_to_lru_tail(bo, mem, NULL); - spin_unlock(&ttm_bo_glob.lru_lock); + spin_unlock(&ttm_glob.lru_lock); return 0; } @@ -839,7 +821,7 @@ int ttm_bo_mem_space(struct ttm_buffer_object *bo, struct ttm_resource *mem, struct ttm_operation_ctx *ctx) { - struct ttm_bo_device *bdev = bo->bdev; + struct ttm_device *bdev = bo->bdev; bool type_found = false; int i, ret; @@ -1057,7 +1039,7 @@ int ttm_bo_validate(struct ttm_buffer_object *bo, } EXPORT_SYMBOL(ttm_bo_validate); -int ttm_bo_init_reserved(struct ttm_bo_device *bdev, +int ttm_bo_init_reserved(struct ttm_device *bdev, struct ttm_buffer_object *bo, size_t size, enum ttm_bo_type type, @@ -1117,7 +1099,7 @@ int ttm_bo_init_reserved(struct ttm_bo_device *bdev, dma_resv_init(&bo->base._resv); drm_vma_node_reset(&bo->base.vma_node); } - atomic_inc(&ttm_bo_glob.bo_count); + atomic_inc(&ttm_glob.bo_count); /* * For ttm_bo_type_device buffers, allocate @@ -1153,7 +1135,7 @@ int ttm_bo_init_reserved(struct ttm_bo_device *bdev, } EXPORT_SYMBOL(ttm_bo_init_reserved); -int ttm_bo_init(struct ttm_bo_device *bdev, +int ttm_bo_init(struct ttm_device *bdev, struct ttm_buffer_object *bo, size_t size, enum ttm_bo_type type, @@ -1181,7 +1163,7 @@ int ttm_bo_init(struct ttm_bo_device *bdev, } EXPORT_SYMBOL(ttm_bo_init); -size_t ttm_bo_dma_acc_size(struct ttm_bo_device *bdev, +size_t ttm_bo_dma_acc_size(struct ttm_device *bdev, unsigned long bo_size, unsigned struct_size) { @@ -1195,148 +1177,13 @@ size_t ttm_bo_dma_acc_size(struct ttm_bo_device *bdev, } EXPORT_SYMBOL(ttm_bo_dma_acc_size); -static void ttm_bo_global_release(void) -{ - struct ttm_bo_global *glob = &ttm_bo_glob; - - mutex_lock(&ttm_global_mutex); - if (--ttm_bo_glob_use_count > 0) - goto out; - - kobject_del(&glob->kobj); - kobject_put(&glob->kobj); - ttm_mem_global_release(&ttm_mem_glob); - __free_page(glob->dummy_read_page); - memset(glob, 0, sizeof(*glob)); -out: - mutex_unlock(&ttm_global_mutex); -} - -static int ttm_bo_global_init(void) -{ - struct ttm_bo_global *glob = &ttm_bo_glob; - int ret = 0; - unsigned i; - - mutex_lock(&ttm_global_mutex); - if (++ttm_bo_glob_use_count > 1) - goto out; - - ret = ttm_mem_global_init(&ttm_mem_glob); - if (ret) - goto out; - - spin_lock_init(&glob->lru_lock); - glob->dummy_read_page = alloc_page(__GFP_ZERO | GFP_DMA32); - - if (unlikely(glob->dummy_read_page == NULL)) { - ret = -ENOMEM; - goto out; - } - - for (i = 0; i < TTM_MAX_BO_PRIORITY; ++i) - INIT_LIST_HEAD(&glob->swap_lru[i]); - INIT_LIST_HEAD(&glob->device_list); - atomic_set(&glob->bo_count, 0); - - debugfs_create_atomic_t("buffer_objects", 0444, ttm_debugfs_root, - &glob->bo_count); -out: - mutex_unlock(&ttm_global_mutex); - return ret; -} - -int ttm_bo_device_release(struct ttm_bo_device *bdev) -{ - struct ttm_bo_global *glob = &ttm_bo_glob; - int ret = 0; - unsigned i; - struct ttm_resource_manager *man; - - man = ttm_manager_type(bdev, TTM_PL_SYSTEM); - ttm_resource_manager_set_used(man, false); - ttm_set_driver_manager(bdev, TTM_PL_SYSTEM, NULL); - - mutex_lock(&ttm_global_mutex); - list_del(&bdev->device_list); - mutex_unlock(&ttm_global_mutex); - - cancel_delayed_work_sync(&bdev->wq); - - if (ttm_bo_delayed_delete(bdev, true)) - pr_debug("Delayed destroy list was clean\n"); - - spin_lock(&glob->lru_lock); - for (i = 0; i < TTM_MAX_BO_PRIORITY; ++i) - if (list_empty(&man->lru[0])) - pr_debug("Swap list %d was clean\n", i); - spin_unlock(&glob->lru_lock); - - ttm_pool_fini(&bdev->pool); - - if (!ret) - ttm_bo_global_release(); - - return ret; -} -EXPORT_SYMBOL(ttm_bo_device_release); - -static void ttm_bo_init_sysman(struct ttm_bo_device *bdev) -{ - struct ttm_resource_manager *man = &bdev->sysman; - - /* - * Initialize the system memory buffer type. - * Other types need to be driver / IOCTL initialized. - */ - man->use_tt = true; - - ttm_resource_manager_init(man, 0); - ttm_set_driver_manager(bdev, TTM_PL_SYSTEM, man); - ttm_resource_manager_set_used(man, true); -} - -int ttm_bo_device_init(struct ttm_bo_device *bdev, - struct ttm_bo_driver *driver, - struct device *dev, - struct address_space *mapping, - struct drm_vma_offset_manager *vma_manager, - bool use_dma_alloc, bool use_dma32) -{ - struct ttm_bo_global *glob = &ttm_bo_glob; - int ret; - - if (WARN_ON(vma_manager == NULL)) - return -EINVAL; - - ret = ttm_bo_global_init(); - if (ret) - return ret; - - bdev->driver = driver; - - ttm_bo_init_sysman(bdev); - ttm_pool_init(&bdev->pool, dev, use_dma_alloc, use_dma32); - - bdev->vma_manager = vma_manager; - INIT_DELAYED_WORK(&bdev->wq, ttm_bo_delayed_workqueue); - INIT_LIST_HEAD(&bdev->ddestroy); - bdev->dev_mapping = mapping; - mutex_lock(&ttm_global_mutex); - list_add_tail(&bdev->device_list, &glob->device_list); - mutex_unlock(&ttm_global_mutex); - - return 0; -} -EXPORT_SYMBOL(ttm_bo_device_init); - /* * buffer object vm functions. */ void ttm_bo_unmap_virtual(struct ttm_buffer_object *bo) { - struct ttm_bo_device *bdev = bo->bdev; + struct ttm_device *bdev = bo->bdev; drm_vma_node_unmap(&bo->base.vma_node, bdev->dev_mapping); ttm_mem_io_free(bdev, &bo->mem); @@ -1374,7 +1221,7 @@ EXPORT_SYMBOL(ttm_bo_wait); */ int ttm_bo_swapout(struct ttm_operation_ctx *ctx) { - struct ttm_bo_global *glob = &ttm_bo_glob; + struct ttm_global *glob = &ttm_glob; struct ttm_buffer_object *bo; int ret = -EBUSY; bool locked; @@ -1452,8 +1299,8 @@ int ttm_bo_swapout(struct ttm_operation_ctx *ctx) * anyone tries to access a ttm page. */ - if (bo->bdev->driver->swap_notify) - bo->bdev->driver->swap_notify(bo); + if (bo->bdev->funcs->swap_notify) + bo->bdev->funcs->swap_notify(bo); ret = ttm_tt_swapout(bo->bdev, bo->ttm); out: @@ -1478,4 +1325,3 @@ void ttm_bo_tt_destroy(struct ttm_buffer_object *bo) ttm_tt_destroy(bo->bdev, bo->ttm); bo->ttm = NULL; } - diff --git a/drivers/gpu/drm/ttm/ttm_bo_util.c b/drivers/gpu/drm/ttm/ttm_bo_util.c index 398d5013fc39..db0f2661d504 100644 --- a/drivers/gpu/drm/ttm/ttm_bo_util.c +++ b/drivers/gpu/drm/ttm/ttm_bo_util.c @@ -46,33 +46,33 @@ struct ttm_transfer_obj { struct ttm_buffer_object *bo; }; -int ttm_mem_io_reserve(struct ttm_bo_device *bdev, +int ttm_mem_io_reserve(struct ttm_device *bdev, struct ttm_resource *mem) { if (mem->bus.offset || mem->bus.addr) return 0; mem->bus.is_iomem = false; - if (!bdev->driver->io_mem_reserve) + if (!bdev->funcs->io_mem_reserve) return 0; - return bdev->driver->io_mem_reserve(bdev, mem); + return bdev->funcs->io_mem_reserve(bdev, mem); } -void ttm_mem_io_free(struct ttm_bo_device *bdev, +void ttm_mem_io_free(struct ttm_device *bdev, struct ttm_resource *mem) { if (!mem->bus.offset && !mem->bus.addr) return; - if (bdev->driver->io_mem_free) - bdev->driver->io_mem_free(bdev, mem); + if (bdev->funcs->io_mem_free) + bdev->funcs->io_mem_free(bdev, mem); mem->bus.offset = 0; mem->bus.addr = NULL; } -static int ttm_resource_ioremap(struct ttm_bo_device *bdev, +static int ttm_resource_ioremap(struct ttm_device *bdev, struct ttm_resource *mem, void **virtual) { @@ -102,7 +102,7 @@ static int ttm_resource_ioremap(struct ttm_bo_device *bdev, return 0; } -static void ttm_resource_iounmap(struct ttm_bo_device *bdev, +static void ttm_resource_iounmap(struct ttm_device *bdev, struct ttm_resource *mem, void *virtual) { @@ -172,7 +172,7 @@ int ttm_bo_move_memcpy(struct ttm_buffer_object *bo, struct ttm_operation_ctx *ctx, struct ttm_resource *new_mem) { - struct ttm_bo_device *bdev = bo->bdev; + struct ttm_device *bdev = bo->bdev; struct ttm_resource_manager *man = ttm_manager_type(bdev, new_mem->mem_type); struct ttm_tt *ttm = bo->ttm; struct ttm_resource *old_mem = &bo->mem; @@ -300,7 +300,7 @@ static int ttm_buffer_object_transfer(struct ttm_buffer_object *bo, * TODO: Explicit member copy would probably be better here. */ - atomic_inc(&ttm_bo_glob.bo_count); + atomic_inc(&ttm_glob.bo_count); INIT_LIST_HEAD(&fbo->base.ddestroy); INIT_LIST_HEAD(&fbo->base.lru); INIT_LIST_HEAD(&fbo->base.swap); @@ -602,7 +602,7 @@ static int ttm_bo_move_to_ghost(struct ttm_buffer_object *bo, static void ttm_bo_move_pipeline_evict(struct ttm_buffer_object *bo, struct dma_fence *fence) { - struct ttm_bo_device *bdev = bo->bdev; + struct ttm_device *bdev = bo->bdev; struct ttm_resource_manager *from = ttm_manager_type(bdev, bo->mem.mem_type); /** @@ -628,7 +628,7 @@ int ttm_bo_move_accel_cleanup(struct ttm_buffer_object *bo, bool pipeline, struct ttm_resource *new_mem) { - struct ttm_bo_device *bdev = bo->bdev; + struct ttm_device *bdev = bo->bdev; struct ttm_resource_manager *from = ttm_manager_type(bdev, bo->mem.mem_type); struct ttm_resource_manager *man = ttm_manager_type(bdev, new_mem->mem_type); int ret = 0; diff --git a/drivers/gpu/drm/ttm/ttm_bo_vm.c b/drivers/gpu/drm/ttm/ttm_bo_vm.c index 6dc96cf66744..b31b18058965 100644 --- a/drivers/gpu/drm/ttm/ttm_bo_vm.c +++ b/drivers/gpu/drm/ttm/ttm_bo_vm.c @@ -95,10 +95,10 @@ out_unlock: static unsigned long ttm_bo_io_mem_pfn(struct ttm_buffer_object *bo, unsigned long page_offset) { - struct ttm_bo_device *bdev = bo->bdev; + struct ttm_device *bdev = bo->bdev; - if (bdev->driver->io_mem_pfn) - return bdev->driver->io_mem_pfn(bo, page_offset); + if (bdev->funcs->io_mem_pfn) + return bdev->funcs->io_mem_pfn(bo, page_offset); return (bo->mem.bus.offset >> PAGE_SHIFT) + page_offset; } @@ -216,7 +216,7 @@ static vm_fault_t ttm_bo_vm_insert_huge(struct vm_fault *vmf, if (page_to_pfn(ttm->pages[page_offset + i]) != pfn + i) goto out_fallback; } - } else if (bo->bdev->driver->io_mem_pfn) { + } else if (bo->bdev->funcs->io_mem_pfn) { for (i = 1; i < fault_page_size; ++i) { if (ttm_bo_io_mem_pfn(bo, page_offset + i) != pfn + i) goto out_fallback; @@ -278,7 +278,7 @@ vm_fault_t ttm_bo_vm_fault_reserved(struct vm_fault *vmf, { struct vm_area_struct *vma = vmf->vma; struct ttm_buffer_object *bo = vma->vm_private_data; - struct ttm_bo_device *bdev = bo->bdev; + struct ttm_device *bdev = bo->bdev; unsigned long page_offset; unsigned long page_last; unsigned long pfn; @@ -488,8 +488,8 @@ int ttm_bo_vm_access(struct vm_area_struct *vma, unsigned long addr, ret = ttm_bo_vm_access_kmap(bo, offset, buf, len, write); break; default: - if (bo->bdev->driver->access_memory) - ret = bo->bdev->driver->access_memory( + if (bo->bdev->funcs->access_memory) + ret = bo->bdev->funcs->access_memory( bo, offset, buf, len, write); else ret = -EIO; @@ -508,7 +508,7 @@ static const struct vm_operations_struct ttm_bo_vm_ops = { .access = ttm_bo_vm_access, }; -static struct ttm_buffer_object *ttm_bo_vm_lookup(struct ttm_bo_device *bdev, +static struct ttm_buffer_object *ttm_bo_vm_lookup(struct ttm_device *bdev, unsigned long offset, unsigned long pages) { @@ -555,9 +555,8 @@ static void ttm_bo_mmap_vma_setup(struct ttm_buffer_object *bo, struct vm_area_s } int ttm_bo_mmap(struct file *filp, struct vm_area_struct *vma, - struct ttm_bo_device *bdev) + struct ttm_device *bdev) { - struct ttm_bo_driver *driver; struct ttm_buffer_object *bo; int ret; @@ -568,12 +567,11 @@ int ttm_bo_mmap(struct file *filp, struct vm_area_struct *vma, if (unlikely(!bo)) return -EINVAL; - driver = bo->bdev->driver; - if (unlikely(!driver->verify_access)) { + if (unlikely(!bo->bdev->funcs->verify_access)) { ret = -EPERM; goto out_unref; } - ret = driver->verify_access(bo, filp); + ret = bo->bdev->funcs->verify_access(bo, filp); if (unlikely(ret != 0)) goto out_unref; diff --git a/drivers/gpu/drm/ttm/ttm_device.c b/drivers/gpu/drm/ttm/ttm_device.c new file mode 100644 index 000000000000..ac0903c9e60a --- /dev/null +++ b/drivers/gpu/drm/ttm/ttm_device.c @@ -0,0 +1,195 @@ +/* SPDX-License-Identifier: GPL-2.0 OR MIT */ + +/* + * Copyright (c) 2006-2009 VMware, Inc., Palo Alto, CA., USA + * Copyright 2020 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Christian König + */ + +#define pr_fmt(fmt) "[TTM DEVICE] " fmt + +#include +#include +#include + +#include "ttm_module.h" + +/** + * ttm_global_mutex - protecting the global state + */ +DEFINE_MUTEX(ttm_global_mutex); +unsigned ttm_glob_use_count; +struct ttm_global ttm_glob; +EXPORT_SYMBOL(ttm_glob); + +static void ttm_global_release(void) +{ + struct ttm_global *glob = &ttm_glob; + + mutex_lock(&ttm_global_mutex); + if (--ttm_glob_use_count > 0) + goto out; + + kobject_del(&glob->kobj); + kobject_put(&glob->kobj); + ttm_mem_global_release(&ttm_mem_glob); + __free_page(glob->dummy_read_page); + memset(glob, 0, sizeof(*glob)); +out: + mutex_unlock(&ttm_global_mutex); +} + +static int ttm_global_init(void) +{ + struct ttm_global *glob = &ttm_glob; + int ret = 0; + unsigned i; + + mutex_lock(&ttm_global_mutex); + if (++ttm_glob_use_count > 1) + goto out; + + ret = ttm_mem_global_init(&ttm_mem_glob); + if (ret) + goto out; + + spin_lock_init(&glob->lru_lock); + glob->dummy_read_page = alloc_page(__GFP_ZERO | GFP_DMA32); + + if (unlikely(glob->dummy_read_page == NULL)) { + ret = -ENOMEM; + goto out; + } + + for (i = 0; i < TTM_MAX_BO_PRIORITY; ++i) + INIT_LIST_HEAD(&glob->swap_lru[i]); + INIT_LIST_HEAD(&glob->device_list); + atomic_set(&glob->bo_count, 0); + + debugfs_create_atomic_t("buffer_objects", 0444, ttm_debugfs_root, + &glob->bo_count); +out: + mutex_unlock(&ttm_global_mutex); + return ret; +} + +static void ttm_init_sysman(struct ttm_device *bdev) +{ + struct ttm_resource_manager *man = &bdev->sysman; + + /* + * Initialize the system memory buffer type. + * Other types need to be driver / IOCTL initialized. + */ + man->use_tt = true; + + ttm_resource_manager_init(man, 0); + ttm_set_driver_manager(bdev, TTM_PL_SYSTEM, man); + ttm_resource_manager_set_used(man, true); +} + +static void ttm_device_delayed_workqueue(struct work_struct *work) +{ + struct ttm_device *bdev = + container_of(work, struct ttm_device, wq.work); + + if (!ttm_bo_delayed_delete(bdev, false)) + schedule_delayed_work(&bdev->wq, + ((HZ / 100) < 1) ? 1 : HZ / 100); +} + +/** + * ttm_device_init + * + * @bdev: A pointer to a struct ttm_device to initialize. + * @funcs: Function table for the device. + * @dev: The core kernel device pointer for DMA mappings and allocations. + * @mapping: The address space to use for this bo. + * @vma_manager: A pointer to a vma manager. + * @use_dma_alloc: If coherent DMA allocation API should be used. + * @use_dma32: If we should use GFP_DMA32 for device memory allocations. + * + * Initializes a struct ttm_device: + * Returns: + * !0: Failure. + */ +int ttm_device_init(struct ttm_device *bdev, struct ttm_device_funcs *funcs, + struct device *dev, struct address_space *mapping, + struct drm_vma_offset_manager *vma_manager, + bool use_dma_alloc, bool use_dma32) +{ + struct ttm_global *glob = &ttm_glob; + int ret; + + if (WARN_ON(vma_manager == NULL)) + return -EINVAL; + + ret = ttm_global_init(); + if (ret) + return ret; + + bdev->funcs = funcs; + + ttm_init_sysman(bdev); + ttm_pool_init(&bdev->pool, dev, use_dma_alloc, use_dma32); + + bdev->vma_manager = vma_manager; + INIT_DELAYED_WORK(&bdev->wq, ttm_device_delayed_workqueue); + INIT_LIST_HEAD(&bdev->ddestroy); + bdev->dev_mapping = mapping; + mutex_lock(&ttm_global_mutex); + list_add_tail(&bdev->device_list, &glob->device_list); + mutex_unlock(&ttm_global_mutex); + + return 0; +} +EXPORT_SYMBOL(ttm_device_init); + +void ttm_device_fini(struct ttm_device *bdev) +{ + struct ttm_global *glob = &ttm_glob; + struct ttm_resource_manager *man; + unsigned i; + + man = ttm_manager_type(bdev, TTM_PL_SYSTEM); + ttm_resource_manager_set_used(man, false); + ttm_set_driver_manager(bdev, TTM_PL_SYSTEM, NULL); + + mutex_lock(&ttm_global_mutex); + list_del(&bdev->device_list); + mutex_unlock(&ttm_global_mutex); + + cancel_delayed_work_sync(&bdev->wq); + + if (ttm_bo_delayed_delete(bdev, true)) + pr_debug("Delayed destroy list was clean\n"); + + spin_lock(&glob->lru_lock); + for (i = 0; i < TTM_MAX_BO_PRIORITY; ++i) + if (list_empty(&man->lru[0])) + pr_debug("Swap list %d was clean\n", i); + spin_unlock(&glob->lru_lock); + + ttm_pool_fini(&bdev->pool); + ttm_global_release(); +} +EXPORT_SYMBOL(ttm_device_fini); diff --git a/drivers/gpu/drm/ttm/ttm_execbuf_util.c b/drivers/gpu/drm/ttm/ttm_execbuf_util.c index 9fa36ed59429..690ab97d52b7 100644 --- a/drivers/gpu/drm/ttm/ttm_execbuf_util.c +++ b/drivers/gpu/drm/ttm/ttm_execbuf_util.c @@ -51,14 +51,14 @@ void ttm_eu_backoff_reservation(struct ww_acquire_ctx *ticket, if (list_empty(list)) return; - spin_lock(&ttm_bo_glob.lru_lock); + spin_lock(&ttm_glob.lru_lock); list_for_each_entry(entry, list, head) { struct ttm_buffer_object *bo = entry->bo; ttm_bo_move_to_lru_tail(bo, &bo->mem, NULL); dma_resv_unlock(bo->base.resv); } - spin_unlock(&ttm_bo_glob.lru_lock); + spin_unlock(&ttm_glob.lru_lock); if (ticket) ww_acquire_fini(ticket); @@ -154,7 +154,7 @@ void ttm_eu_fence_buffer_objects(struct ww_acquire_ctx *ticket, if (list_empty(list)) return; - spin_lock(&ttm_bo_glob.lru_lock); + spin_lock(&ttm_glob.lru_lock); list_for_each_entry(entry, list, head) { struct ttm_buffer_object *bo = entry->bo; @@ -165,7 +165,7 @@ void ttm_eu_fence_buffer_objects(struct ww_acquire_ctx *ticket, ttm_bo_move_to_lru_tail(bo, &bo->mem, NULL); dma_resv_unlock(bo->base.resv); } - spin_unlock(&ttm_bo_glob.lru_lock); + spin_unlock(&ttm_glob.lru_lock); if (ticket) ww_acquire_fini(ticket); } diff --git a/drivers/gpu/drm/ttm/ttm_range_manager.c b/drivers/gpu/drm/ttm/ttm_range_manager.c index a39305f742da..707e5c152896 100644 --- a/drivers/gpu/drm/ttm/ttm_range_manager.c +++ b/drivers/gpu/drm/ttm/ttm_range_manager.c @@ -111,7 +111,7 @@ static void ttm_range_man_free(struct ttm_resource_manager *man, static const struct ttm_resource_manager_func ttm_range_manager_func; -int ttm_range_man_init(struct ttm_bo_device *bdev, +int ttm_range_man_init(struct ttm_device *bdev, unsigned type, bool use_tt, unsigned long p_size) { @@ -138,7 +138,7 @@ int ttm_range_man_init(struct ttm_bo_device *bdev, } EXPORT_SYMBOL(ttm_range_man_init); -int ttm_range_man_fini(struct ttm_bo_device *bdev, +int ttm_range_man_fini(struct ttm_device *bdev, unsigned type) { struct ttm_resource_manager *man = ttm_manager_type(bdev, type); diff --git a/drivers/gpu/drm/ttm/ttm_resource.c b/drivers/gpu/drm/ttm/ttm_resource.c index b60699bf4816..ed1672a9f332 100644 --- a/drivers/gpu/drm/ttm/ttm_resource.c +++ b/drivers/gpu/drm/ttm/ttm_resource.c @@ -83,7 +83,7 @@ EXPORT_SYMBOL(ttm_resource_manager_init); * Evict all the objects out of a memory manager until it is empty. * Part of memory manager cleanup sequence. */ -int ttm_resource_manager_evict_all(struct ttm_bo_device *bdev, +int ttm_resource_manager_evict_all(struct ttm_device *bdev, struct ttm_resource_manager *man) { struct ttm_operation_ctx ctx = { @@ -91,7 +91,7 @@ int ttm_resource_manager_evict_all(struct ttm_bo_device *bdev, .no_wait_gpu = false, .force_alloc = true }; - struct ttm_bo_global *glob = &ttm_bo_glob; + struct ttm_global *glob = &ttm_glob; struct dma_fence *fence; int ret; unsigned i; diff --git a/drivers/gpu/drm/ttm/ttm_tt.c b/drivers/gpu/drm/ttm/ttm_tt.c index 7f75a13163f0..7782d5393c7c 100644 --- a/drivers/gpu/drm/ttm/ttm_tt.c +++ b/drivers/gpu/drm/ttm/ttm_tt.c @@ -43,7 +43,7 @@ */ int ttm_tt_create(struct ttm_buffer_object *bo, bool zero_alloc) { - struct ttm_bo_device *bdev = bo->bdev; + struct ttm_device *bdev = bo->bdev; uint32_t page_flags = 0; dma_resv_assert_held(bo->base.resv); @@ -66,7 +66,7 @@ int ttm_tt_create(struct ttm_buffer_object *bo, bool zero_alloc) return -EINVAL; } - bo->ttm = bdev->driver->ttm_tt_create(bo, page_flags); + bo->ttm = bdev->funcs->ttm_tt_create(bo, page_flags); if (unlikely(bo->ttm == NULL)) return -ENOMEM; @@ -108,7 +108,7 @@ static int ttm_sg_tt_alloc_page_directory(struct ttm_tt *ttm) return 0; } -void ttm_tt_destroy_common(struct ttm_bo_device *bdev, struct ttm_tt *ttm) +void ttm_tt_destroy_common(struct ttm_device *bdev, struct ttm_tt *ttm) { ttm_tt_unpopulate(bdev, ttm); @@ -119,9 +119,9 @@ void ttm_tt_destroy_common(struct ttm_bo_device *bdev, struct ttm_tt *ttm) } EXPORT_SYMBOL(ttm_tt_destroy_common); -void ttm_tt_destroy(struct ttm_bo_device *bdev, struct ttm_tt *ttm) +void ttm_tt_destroy(struct ttm_device *bdev, struct ttm_tt *ttm) { - bdev->driver->ttm_tt_destroy(bdev, ttm); + bdev->funcs->ttm_tt_destroy(bdev, ttm); } static void ttm_tt_init_fields(struct ttm_tt *ttm, @@ -223,7 +223,7 @@ out_err: return ret; } -int ttm_tt_swapout(struct ttm_bo_device *bdev, struct ttm_tt *ttm) +int ttm_tt_swapout(struct ttm_device *bdev, struct ttm_tt *ttm) { struct address_space *swap_space; struct file *swap_storage; @@ -271,7 +271,7 @@ out_err: return ret; } -static void ttm_tt_add_mapping(struct ttm_bo_device *bdev, struct ttm_tt *ttm) +static void ttm_tt_add_mapping(struct ttm_device *bdev, struct ttm_tt *ttm) { pgoff_t i; @@ -282,7 +282,7 @@ static void ttm_tt_add_mapping(struct ttm_bo_device *bdev, struct ttm_tt *ttm) ttm->pages[i]->mapping = bdev->dev_mapping; } -int ttm_tt_populate(struct ttm_bo_device *bdev, +int ttm_tt_populate(struct ttm_device *bdev, struct ttm_tt *ttm, struct ttm_operation_ctx *ctx) { int ret; @@ -293,8 +293,8 @@ int ttm_tt_populate(struct ttm_bo_device *bdev, if (ttm_tt_is_populated(ttm)) return 0; - if (bdev->driver->ttm_tt_populate) - ret = bdev->driver->ttm_tt_populate(bdev, ttm, ctx); + if (bdev->funcs->ttm_tt_populate) + ret = bdev->funcs->ttm_tt_populate(bdev, ttm, ctx); else ret = ttm_pool_alloc(&bdev->pool, ttm, ctx); if (ret) @@ -328,15 +328,15 @@ static void ttm_tt_clear_mapping(struct ttm_tt *ttm) } } -void ttm_tt_unpopulate(struct ttm_bo_device *bdev, +void ttm_tt_unpopulate(struct ttm_device *bdev, struct ttm_tt *ttm) { if (!ttm_tt_is_populated(ttm)) return; ttm_tt_clear_mapping(ttm); - if (bdev->driver->ttm_tt_unpopulate) - bdev->driver->ttm_tt_unpopulate(bdev, ttm); + if (bdev->funcs->ttm_tt_unpopulate) + bdev->funcs->ttm_tt_unpopulate(bdev, ttm); else ttm_pool_free(&bdev->pool, ttm); ttm->page_flags &= ~TTM_PAGE_FLAG_PRIV_POPULATED; diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_blit.c b/drivers/gpu/drm/vmwgfx/vmwgfx_blit.c index 118db24eb756..3a438ae4d3f4 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_blit.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_blit.c @@ -466,13 +466,13 @@ int vmw_bo_cpu_blit(struct ttm_buffer_object *dst, dma_resv_assert_held(src->base.resv); if (!ttm_tt_is_populated(dst->ttm)) { - ret = dst->bdev->driver->ttm_tt_populate(dst->bdev, dst->ttm, &ctx); + ret = dst->bdev->funcs->ttm_tt_populate(dst->bdev, dst->ttm, &ctx); if (ret) return ret; } if (!ttm_tt_is_populated(src->ttm)) { - ret = src->bdev->driver->ttm_tt_populate(src->bdev, src->ttm, &ctx); + ret = src->bdev->funcs->ttm_tt_populate(src->bdev, src->ttm, &ctx); if (ret) return ret; } diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_bo.c b/drivers/gpu/drm/vmwgfx/vmwgfx_bo.c index ab0844b47d4d..6b3bfd8c678a 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_bo.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_bo.c @@ -545,7 +545,7 @@ int vmw_bo_init(struct vmw_private *dev_priv, void (*bo_free)(struct ttm_buffer_object *bo)) { struct ttm_operation_ctx ctx = { interruptible, false }; - struct ttm_bo_device *bdev = &dev_priv->bdev; + struct ttm_device *bdev = &dev_priv->bdev; size_t acc_size; int ret; bool user = (bo_free == &vmw_user_bo_destroy); @@ -1058,7 +1058,7 @@ int vmw_user_bo_reference(struct ttm_object_file *tfile, void vmw_bo_fence_single(struct ttm_buffer_object *bo, struct vmw_fence_obj *fence) { - struct ttm_bo_device *bdev = bo->bdev; + struct ttm_device *bdev = bo->bdev; struct vmw_private *dev_priv = container_of(bdev, struct vmw_private, bdev); diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c index eb997f4678de..b454d80c273e 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c @@ -884,12 +884,12 @@ static int vmw_driver_load(struct vmw_private *dev_priv, u32 pci_id) drm_vma_offset_manager_init(&dev_priv->vma_manager, DRM_FILE_PAGE_OFFSET_START, DRM_FILE_PAGE_OFFSET_SIZE); - ret = ttm_bo_device_init(&dev_priv->bdev, &vmw_bo_driver, - dev_priv->drm.dev, - dev_priv->drm.anon_inode->i_mapping, - &dev_priv->vma_manager, - dev_priv->map_mode == vmw_dma_alloc_coherent, - false); + ret = ttm_device_init(&dev_priv->bdev, &vmw_bo_driver, + dev_priv->drm.dev, + dev_priv->drm.anon_inode->i_mapping, + &dev_priv->vma_manager, + dev_priv->map_mode == vmw_dma_alloc_coherent, + false); if (unlikely(ret != 0)) { DRM_ERROR("Failed initializing TTM buffer object driver.\n"); goto out_no_bdev; @@ -1006,7 +1006,7 @@ out_no_kms: vmw_gmrid_man_fini(dev_priv, VMW_PL_GMR); vmw_vram_manager_fini(dev_priv); out_no_vram: - (void)ttm_bo_device_release(&dev_priv->bdev); + ttm_device_fini(&dev_priv->bdev); out_no_bdev: vmw_fence_manager_takedown(dev_priv->fman); out_no_fman: @@ -1053,7 +1053,7 @@ static void vmw_driver_unload(struct drm_device *dev) if (dev_priv->has_mob) vmw_gmrid_man_fini(dev_priv, VMW_PL_MOB); vmw_vram_manager_fini(dev_priv); - (void) ttm_bo_device_release(&dev_priv->bdev); + ttm_device_fini(&dev_priv->bdev); drm_vma_offset_manager_destroy(&dev_priv->vma_manager); vmw_release_device_late(dev_priv); vmw_fence_manager_takedown(dev_priv->fman); diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h index 336f5086ca26..e65b00f8336d 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h @@ -484,7 +484,7 @@ enum vmw_sm_type { struct vmw_private { struct drm_device drm; - struct ttm_bo_device bdev; + struct ttm_device bdev; struct vmw_fifo_state fifo; @@ -999,7 +999,7 @@ extern struct ttm_placement vmw_evictable_placement; extern struct ttm_placement vmw_srf_placement; extern struct ttm_placement vmw_mob_placement; extern struct ttm_placement vmw_nonfixed_placement; -extern struct ttm_bo_driver vmw_bo_driver; +extern struct ttm_device_funcs vmw_bo_driver; extern const struct vmw_sg_table * vmw_bo_sg_table(struct ttm_buffer_object *bo); extern int vmw_bo_create_and_populate(struct vmw_private *dev_priv, diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c b/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c index c4df51a2a926..c3a724e37104 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c @@ -856,7 +856,7 @@ void vmw_query_move_notify(struct ttm_buffer_object *bo, struct ttm_resource *mem) { struct vmw_buffer_object *dx_query_mob; - struct ttm_bo_device *bdev = bo->bdev; + struct ttm_device *bdev = bo->bdev; struct vmw_private *dev_priv; diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_ttm_buffer.c b/drivers/gpu/drm/vmwgfx/vmwgfx_ttm_buffer.c index 6a44567e4ba5..d1bfa59579f1 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_ttm_buffer.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_ttm_buffer.c @@ -483,7 +483,7 @@ const struct vmw_sg_table *vmw_bo_sg_table(struct ttm_buffer_object *bo) } -static int vmw_ttm_bind(struct ttm_bo_device *bdev, +static int vmw_ttm_bind(struct ttm_device *bdev, struct ttm_tt *ttm, struct ttm_resource *bo_mem) { struct vmw_ttm_tt *vmw_be = @@ -527,7 +527,7 @@ static int vmw_ttm_bind(struct ttm_bo_device *bdev, return ret; } -static void vmw_ttm_unbind(struct ttm_bo_device *bdev, +static void vmw_ttm_unbind(struct ttm_device *bdev, struct ttm_tt *ttm) { struct vmw_ttm_tt *vmw_be = @@ -553,7 +553,7 @@ static void vmw_ttm_unbind(struct ttm_bo_device *bdev, } -static void vmw_ttm_destroy(struct ttm_bo_device *bdev, struct ttm_tt *ttm) +static void vmw_ttm_destroy(struct ttm_device *bdev, struct ttm_tt *ttm) { struct vmw_ttm_tt *vmw_be = container_of(ttm, struct vmw_ttm_tt, dma_ttm); @@ -573,7 +573,7 @@ static void vmw_ttm_destroy(struct ttm_bo_device *bdev, struct ttm_tt *ttm) } -static int vmw_ttm_populate(struct ttm_bo_device *bdev, +static int vmw_ttm_populate(struct ttm_device *bdev, struct ttm_tt *ttm, struct ttm_operation_ctx *ctx) { /* TODO: maybe completely drop this ? */ @@ -583,7 +583,7 @@ static int vmw_ttm_populate(struct ttm_bo_device *bdev, return ttm_pool_alloc(&bdev->pool, ttm, ctx); } -static void vmw_ttm_unpopulate(struct ttm_bo_device *bdev, +static void vmw_ttm_unpopulate(struct ttm_device *bdev, struct ttm_tt *ttm) { struct vmw_ttm_tt *vmw_tt = container_of(ttm, struct vmw_ttm_tt, @@ -640,7 +640,7 @@ static int vmw_verify_access(struct ttm_buffer_object *bo, struct file *filp) return vmw_user_bo_verify_access(bo, tfile); } -static int vmw_ttm_io_mem_reserve(struct ttm_bo_device *bdev, struct ttm_resource *mem) +static int vmw_ttm_io_mem_reserve(struct ttm_device *bdev, struct ttm_resource *mem) { struct vmw_private *dev_priv = container_of(bdev, struct vmw_private, bdev); @@ -744,7 +744,7 @@ vmw_delete_mem_notify(struct ttm_buffer_object *bo) vmw_move_notify(bo, false, NULL); } -struct ttm_bo_driver vmw_bo_driver = { +struct ttm_device_funcs vmw_bo_driver = { .ttm_tt_create = &vmw_ttm_tt_create, .ttm_tt_populate = &vmw_ttm_populate, .ttm_tt_unpopulate = &vmw_ttm_unpopulate, diff --git a/include/drm/drm_gem_vram_helper.h b/include/drm/drm_gem_vram_helper.h index a4bac02249c2..288055d397d9 100644 --- a/include/drm/drm_gem_vram_helper.h +++ b/include/drm/drm_gem_vram_helper.h @@ -172,19 +172,19 @@ struct drm_vram_mm { uint64_t vram_base; size_t vram_size; - struct ttm_bo_device bdev; + struct ttm_device bdev; }; /** * drm_vram_mm_of_bdev() - \ - Returns the container of type &struct ttm_bo_device for field bdev. + Returns the container of type &struct ttm_device for field bdev. * @bdev: the TTM BO device * * Returns: * The containing instance of &struct drm_vram_mm */ static inline struct drm_vram_mm *drm_vram_mm_of_bdev( - struct ttm_bo_device *bdev) + struct ttm_device *bdev) { return container_of(bdev, struct drm_vram_mm, bdev); } diff --git a/include/drm/ttm/ttm_bo_api.h b/include/drm/ttm/ttm_bo_api.h index e17be324d95f..62734db0b421 100644 --- a/include/drm/ttm/ttm_bo_api.h +++ b/include/drm/ttm/ttm_bo_api.h @@ -44,9 +44,9 @@ #include "ttm_resource.h" -struct ttm_bo_global; +struct ttm_global; -struct ttm_bo_device; +struct ttm_device; struct dma_buf_map; @@ -122,7 +122,7 @@ struct ttm_buffer_object { * Members constant at init. */ - struct ttm_bo_device *bdev; + struct ttm_device *bdev; enum ttm_bo_type type; void (*destroy) (struct ttm_buffer_object *); size_t acc_size; @@ -313,7 +313,7 @@ void ttm_bo_put(struct ttm_buffer_object *bo); * @bulk: optional bulk move structure to remember BO positions * * Move this BO to the tail of all lru lists used to lookup and reserve an - * object. This function must be called with struct ttm_bo_global::lru_lock + * object. This function must be called with struct ttm_global::lru_lock * held, and is used to make a BO less likely to be considered for eviction. */ void ttm_bo_move_to_lru_tail(struct ttm_buffer_object *bo, @@ -326,7 +326,7 @@ void ttm_bo_move_to_lru_tail(struct ttm_buffer_object *bo, * @bulk: bulk move structure * * Bulk move BOs to the LRU tail, only valid to use when driver makes sure that - * BO order never changes. Should be called with ttm_bo_global::lru_lock held. + * BO order never changes. Should be called with ttm_global::lru_lock held. */ void ttm_bo_bulk_move_lru_tail(struct ttm_lru_bulk_move *bulk); @@ -337,14 +337,14 @@ void ttm_bo_bulk_move_lru_tail(struct ttm_lru_bulk_move *bulk); * Returns * True if the workqueue was queued at the time */ -int ttm_bo_lock_delayed_workqueue(struct ttm_bo_device *bdev); +int ttm_bo_lock_delayed_workqueue(struct ttm_device *bdev); /** * ttm_bo_unlock_delayed_workqueue * * Allows the delayed workqueue to run. */ -void ttm_bo_unlock_delayed_workqueue(struct ttm_bo_device *bdev, int resched); +void ttm_bo_unlock_delayed_workqueue(struct ttm_device *bdev, int resched); /** * ttm_bo_eviction_valuable @@ -357,14 +357,14 @@ void ttm_bo_unlock_delayed_workqueue(struct ttm_bo_device *bdev, int resched); bool ttm_bo_eviction_valuable(struct ttm_buffer_object *bo, const struct ttm_place *place); -size_t ttm_bo_dma_acc_size(struct ttm_bo_device *bdev, +size_t ttm_bo_dma_acc_size(struct ttm_device *bdev, unsigned long bo_size, unsigned struct_size); /** * ttm_bo_init_reserved * - * @bdev: Pointer to a ttm_bo_device struct. + * @bdev: Pointer to a ttm_device struct. * @bo: Pointer to a ttm_buffer_object to be initialized. * @size: Requested size of buffer object. * @type: Requested type of buffer object. @@ -396,7 +396,7 @@ size_t ttm_bo_dma_acc_size(struct ttm_bo_device *bdev, * -ERESTARTSYS: Interrupted by signal while sleeping waiting for resources. */ -int ttm_bo_init_reserved(struct ttm_bo_device *bdev, +int ttm_bo_init_reserved(struct ttm_device *bdev, struct ttm_buffer_object *bo, size_t size, enum ttm_bo_type type, struct ttm_placement *placement, @@ -409,7 +409,7 @@ int ttm_bo_init_reserved(struct ttm_bo_device *bdev, /** * ttm_bo_init * - * @bdev: Pointer to a ttm_bo_device struct. + * @bdev: Pointer to a ttm_device struct. * @bo: Pointer to a ttm_buffer_object to be initialized. * @size: Requested size of buffer object. * @type: Requested type of buffer object. @@ -443,7 +443,7 @@ int ttm_bo_init_reserved(struct ttm_bo_device *bdev, * -EINVAL: Invalid placement flags. * -ERESTARTSYS: Interrupted by signal while sleeping waiting for resources. */ -int ttm_bo_init(struct ttm_bo_device *bdev, struct ttm_buffer_object *bo, +int ttm_bo_init(struct ttm_device *bdev, struct ttm_buffer_object *bo, size_t size, enum ttm_bo_type type, struct ttm_placement *placement, uint32_t page_alignment, bool interrubtible, size_t acc_size, @@ -537,18 +537,18 @@ int ttm_bo_mmap_obj(struct vm_area_struct *vma, struct ttm_buffer_object *bo); * * @filp: filp as input from the mmap method. * @vma: vma as input from the mmap method. - * @bdev: Pointer to the ttm_bo_device with the address space manager. + * @bdev: Pointer to the ttm_device with the address space manager. * * This function is intended to be called by the device mmap method. * if the device address space is to be backed by the bo manager. */ int ttm_bo_mmap(struct file *filp, struct vm_area_struct *vma, - struct ttm_bo_device *bdev); + struct ttm_device *bdev); /** * ttm_bo_io * - * @bdev: Pointer to the struct ttm_bo_device. + * @bdev: Pointer to the struct ttm_device. * @filp: Pointer to the struct file attempting to read / write. * @wbuf: User-space pointer to address of buffer to write. NULL on read. * @rbuf: User-space pointer to address of buffer to read into. @@ -565,7 +565,7 @@ int ttm_bo_mmap(struct file *filp, struct vm_area_struct *vma, * the function may return -ERESTARTSYS if * interrupted by a signal. */ -ssize_t ttm_bo_io(struct ttm_bo_device *bdev, struct file *filp, +ssize_t ttm_bo_io(struct ttm_device *bdev, struct file *filp, const char __user *wbuf, char __user *rbuf, size_t count, loff_t *f_pos, bool write); @@ -617,7 +617,7 @@ static inline void ttm_bo_unpin(struct ttm_buffer_object *bo) --bo->pin_count; } -int ttm_mem_evict_first(struct ttm_bo_device *bdev, +int ttm_mem_evict_first(struct ttm_device *bdev, struct ttm_resource_manager *man, const struct ttm_place *place, struct ttm_operation_ctx *ctx, @@ -642,5 +642,6 @@ void ttm_bo_vm_close(struct vm_area_struct *vma); int ttm_bo_vm_access(struct vm_area_struct *vma, unsigned long addr, void *buf, int len, int write); +bool ttm_bo_delayed_delete(struct ttm_device *bdev, bool remove_all); #endif diff --git a/include/drm/ttm/ttm_bo_driver.h b/include/drm/ttm/ttm_bo_driver.h index 423348414c59..1c9bf993e252 100644 --- a/include/drm/ttm/ttm_bo_driver.h +++ b/include/drm/ttm/ttm_bo_driver.h @@ -37,301 +37,14 @@ #include #include +#include + #include "ttm_bo_api.h" #include "ttm_memory.h" #include "ttm_placement.h" #include "ttm_tt.h" #include "ttm_pool.h" -/** - * struct ttm_bo_driver - * - * @create_ttm_backend_entry: Callback to create a struct ttm_backend. - * @evict_flags: Callback to obtain placement flags when a buffer is evicted. - * @move: Callback for a driver to hook in accelerated functions to - * move a buffer. - * If set to NULL, a potentially slow memcpy() move is used. - */ - -struct ttm_bo_driver { - /** - * ttm_tt_create - * - * @bo: The buffer object to create the ttm for. - * @page_flags: Page flags as identified by TTM_PAGE_FLAG_XX flags. - * - * Create a struct ttm_tt to back data with system memory pages. - * No pages are actually allocated. - * Returns: - * NULL: Out of memory. - */ - struct ttm_tt *(*ttm_tt_create)(struct ttm_buffer_object *bo, - uint32_t page_flags); - - /** - * ttm_tt_populate - * - * @ttm: The struct ttm_tt to contain the backing pages. - * - * Allocate all backing pages - * Returns: - * -ENOMEM: Out of memory. - */ - int (*ttm_tt_populate)(struct ttm_bo_device *bdev, - struct ttm_tt *ttm, - struct ttm_operation_ctx *ctx); - - /** - * ttm_tt_unpopulate - * - * @ttm: The struct ttm_tt to contain the backing pages. - * - * Free all backing page - */ - void (*ttm_tt_unpopulate)(struct ttm_bo_device *bdev, struct ttm_tt *ttm); - - /** - * ttm_tt_destroy - * - * @bdev: Pointer to a ttm device - * @ttm: Pointer to a struct ttm_tt. - * - * Destroy the backend. This will be call back from ttm_tt_destroy so - * don't call ttm_tt_destroy from the callback or infinite loop. - */ - void (*ttm_tt_destroy)(struct ttm_bo_device *bdev, struct ttm_tt *ttm); - - /** - * struct ttm_bo_driver member eviction_valuable - * - * @bo: the buffer object to be evicted - * @place: placement we need room for - * - * Check with the driver if it is valuable to evict a BO to make room - * for a certain placement. - */ - bool (*eviction_valuable)(struct ttm_buffer_object *bo, - const struct ttm_place *place); - /** - * struct ttm_bo_driver member evict_flags: - * - * @bo: the buffer object to be evicted - * - * Return the bo flags for a buffer which is not mapped to the hardware. - * These will be placed in proposed_flags so that when the move is - * finished, they'll end up in bo->mem.flags - * This should not cause multihop evictions, and the core will warn - * if one is proposed. - */ - - void (*evict_flags)(struct ttm_buffer_object *bo, - struct ttm_placement *placement); - - /** - * struct ttm_bo_driver member move: - * - * @bo: the buffer to move - * @evict: whether this motion is evicting the buffer from - * the graphics address space - * @ctx: context for this move with parameters - * @new_mem: the new memory region receiving the buffer - @ @hop: placement for driver directed intermediate hop - * - * Move a buffer between two memory regions. - * Returns errno -EMULTIHOP if driver requests a hop - */ - int (*move)(struct ttm_buffer_object *bo, bool evict, - struct ttm_operation_ctx *ctx, - struct ttm_resource *new_mem, - struct ttm_place *hop); - - /** - * struct ttm_bo_driver_member verify_access - * - * @bo: Pointer to a buffer object. - * @filp: Pointer to a struct file trying to access the object. - * - * Called from the map / write / read methods to verify that the - * caller is permitted to access the buffer object. - * This member may be set to NULL, which will refuse this kind of - * access for all buffer objects. - * This function should return 0 if access is granted, -EPERM otherwise. - */ - int (*verify_access)(struct ttm_buffer_object *bo, - struct file *filp); - - /** - * Hook to notify driver about a resource delete. - */ - void (*delete_mem_notify)(struct ttm_buffer_object *bo); - - /** - * notify the driver that we're about to swap out this bo - */ - void (*swap_notify)(struct ttm_buffer_object *bo); - - /** - * Driver callback on when mapping io memory (for bo_move_memcpy - * for instance). TTM will take care to call io_mem_free whenever - * the mapping is not use anymore. io_mem_reserve & io_mem_free - * are balanced. - */ - int (*io_mem_reserve)(struct ttm_bo_device *bdev, - struct ttm_resource *mem); - void (*io_mem_free)(struct ttm_bo_device *bdev, - struct ttm_resource *mem); - - /** - * Return the pfn for a given page_offset inside the BO. - * - * @bo: the BO to look up the pfn for - * @page_offset: the offset to look up - */ - unsigned long (*io_mem_pfn)(struct ttm_buffer_object *bo, - unsigned long page_offset); - - /** - * Read/write memory buffers for ptrace access - * - * @bo: the BO to access - * @offset: the offset from the start of the BO - * @buf: pointer to source/destination buffer - * @len: number of bytes to copy - * @write: whether to read (0) from or write (non-0) to BO - * - * If successful, this function should return the number of - * bytes copied, -EIO otherwise. If the number of bytes - * returned is < len, the function may be called again with - * the remainder of the buffer to copy. - */ - int (*access_memory)(struct ttm_buffer_object *bo, unsigned long offset, - void *buf, int len, int write); - - /** - * struct ttm_bo_driver member del_from_lru_notify - * - * @bo: the buffer object deleted from lru - * - * notify driver that a BO was deleted from LRU. - */ - void (*del_from_lru_notify)(struct ttm_buffer_object *bo); - - /** - * Notify the driver that we're about to release a BO - * - * @bo: BO that is about to be released - * - * Gives the driver a chance to do any cleanup, including - * adding fences that may force a delayed delete - */ - void (*release_notify)(struct ttm_buffer_object *bo); -}; - -/** - * struct ttm_bo_global - Buffer object driver global data. - * - * @dummy_read_page: Pointer to a dummy page used for mapping requests - * of unpopulated pages. - * @shrink: A shrink callback object used for buffer object swap. - * @device_list_mutex: Mutex protecting the device list. - * This mutex is held while traversing the device list for pm options. - * @lru_lock: Spinlock protecting the bo subsystem lru lists. - * @device_list: List of buffer object devices. - * @swap_lru: Lru list of buffer objects used for swapping. - */ - -extern struct ttm_bo_global { - - /** - * Constant after init. - */ - - struct kobject kobj; - struct page *dummy_read_page; - spinlock_t lru_lock; - - /** - * Protected by ttm_global_mutex. - */ - struct list_head device_list; - - /** - * Protected by the lru_lock. - */ - struct list_head swap_lru[TTM_MAX_BO_PRIORITY]; - - /** - * Internal protection. - */ - atomic_t bo_count; -} ttm_bo_glob; - - -#define TTM_NUM_MEM_TYPES 8 - -/** - * struct ttm_bo_device - Buffer object driver device-specific data. - * - * @driver: Pointer to a struct ttm_bo_driver struct setup by the driver. - * @man: An array of resource_managers. - * @vma_manager: Address space manager (pointer) - * lru_lock: Spinlock that protects the buffer+device lru lists and - * ddestroy lists. - * @dev_mapping: A pointer to the struct address_space representing the - * device address space. - * @wq: Work queue structure for the delayed delete workqueue. - * - */ - -struct ttm_bo_device { - - /* - * Constant after bo device init / atomic. - */ - struct list_head device_list; - struct ttm_bo_driver *driver; - /* - * access via ttm_manager_type. - */ - struct ttm_resource_manager sysman; - struct ttm_resource_manager *man_drv[TTM_NUM_MEM_TYPES]; - /* - * Protected by internal locks. - */ - struct drm_vma_offset_manager *vma_manager; - struct ttm_pool pool; - - /* - * Protected by the global:lru lock. - */ - struct list_head ddestroy; - - /* - * Protected by load / firstopen / lastclose /unload sync. - */ - - struct address_space *dev_mapping; - - /* - * Internal protection. - */ - - struct delayed_work wq; -}; - -static inline struct ttm_resource_manager *ttm_manager_type(struct ttm_bo_device *bdev, - int mem_type) -{ - return bdev->man_drv[mem_type]; -} - -static inline void ttm_set_driver_manager(struct ttm_bo_device *bdev, - int type, - struct ttm_resource_manager *manager) -{ - bdev->man_drv[type] = manager; -} - /** * struct ttm_lru_bulk_move_pos * @@ -388,31 +101,6 @@ int ttm_bo_mem_space(struct ttm_buffer_object *bo, struct ttm_resource *mem, struct ttm_operation_ctx *ctx); -int ttm_bo_device_release(struct ttm_bo_device *bdev); - -/** - * ttm_bo_device_init - * - * @bdev: A pointer to a struct ttm_bo_device to initialize. - * @glob: A pointer to an initialized struct ttm_bo_global. - * @driver: A pointer to a struct ttm_bo_driver set up by the caller. - * @dev: The core kernel device pointer for DMA mappings and allocations. - * @mapping: The address space to use for this bo. - * @vma_manager: A pointer to a vma manager. - * @use_dma_alloc: If coherent DMA allocation API should be used. - * @use_dma32: If we should use GFP_DMA32 for device memory allocations. - * - * Initializes a struct ttm_bo_device: - * Returns: - * !0: Failure. - */ -int ttm_bo_device_init(struct ttm_bo_device *bdev, - struct ttm_bo_driver *driver, - struct device *dev, - struct address_space *mapping, - struct drm_vma_offset_manager *vma_manager, - bool use_dma_alloc, bool use_dma32); - /** * ttm_bo_unmap_virtual * @@ -494,9 +182,9 @@ static inline int ttm_bo_reserve_slowpath(struct ttm_buffer_object *bo, static inline void ttm_bo_move_to_lru_tail_unlocked(struct ttm_buffer_object *bo) { - spin_lock(&ttm_bo_glob.lru_lock); + spin_lock(&ttm_glob.lru_lock); ttm_bo_move_to_lru_tail(bo, &bo->mem, NULL); - spin_unlock(&ttm_bo_glob.lru_lock); + spin_unlock(&ttm_glob.lru_lock); } static inline void ttm_bo_assign_mem(struct ttm_buffer_object *bo, @@ -538,9 +226,9 @@ static inline void ttm_bo_unreserve(struct ttm_buffer_object *bo) /* * ttm_bo_util.c */ -int ttm_mem_io_reserve(struct ttm_bo_device *bdev, +int ttm_mem_io_reserve(struct ttm_device *bdev, struct ttm_resource *mem); -void ttm_mem_io_free(struct ttm_bo_device *bdev, +void ttm_mem_io_free(struct ttm_device *bdev, struct ttm_resource *mem); /** @@ -631,7 +319,7 @@ void ttm_bo_tt_destroy(struct ttm_buffer_object *bo); * Initialise a generic range manager for the selected memory type. * The range manager is installed for this device in the type slot. */ -int ttm_range_man_init(struct ttm_bo_device *bdev, +int ttm_range_man_init(struct ttm_device *bdev, unsigned type, bool use_tt, unsigned long p_size); @@ -643,7 +331,7 @@ int ttm_range_man_init(struct ttm_bo_device *bdev, * * Remove the generic range manager from a slot and tear it down. */ -int ttm_range_man_fini(struct ttm_bo_device *bdev, +int ttm_range_man_fini(struct ttm_device *bdev, unsigned type); #endif diff --git a/include/drm/ttm/ttm_device.h b/include/drm/ttm/ttm_device.h new file mode 100644 index 000000000000..7bc8bb797161 --- /dev/null +++ b/include/drm/ttm/ttm_device.h @@ -0,0 +1,319 @@ +/* + * Copyright 2020 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: Christian König + */ + +#ifndef _TTM_DEVICE_H_ +#define _TTM_DEVICE_H_ + +#include +#include +#include +#include + +#define TTM_NUM_MEM_TYPES 8 + +struct ttm_device; +struct ttm_placement; +struct ttm_buffer_object; +struct ttm_operation_ctx; + +/** + * struct ttm_global - Buffer object driver global data. + * + * @dummy_read_page: Pointer to a dummy page used for mapping requests + * of unpopulated pages. + * @shrink: A shrink callback object used for buffer object swap. + * @device_list_mutex: Mutex protecting the device list. + * This mutex is held while traversing the device list for pm options. + * @lru_lock: Spinlock protecting the bo subsystem lru lists. + * @device_list: List of buffer object devices. + * @swap_lru: Lru list of buffer objects used for swapping. + */ +extern struct ttm_global { + + /** + * Constant after init. + */ + + struct kobject kobj; + struct page *dummy_read_page; + spinlock_t lru_lock; + + /** + * Protected by ttm_global_mutex. + */ + struct list_head device_list; + + /** + * Protected by the lru_lock. + */ + struct list_head swap_lru[TTM_MAX_BO_PRIORITY]; + + /** + * Internal protection. + */ + atomic_t bo_count; +} ttm_glob; + +struct ttm_device_funcs { + /** + * ttm_tt_create + * + * @bo: The buffer object to create the ttm for. + * @page_flags: Page flags as identified by TTM_PAGE_FLAG_XX flags. + * + * Create a struct ttm_tt to back data with system memory pages. + * No pages are actually allocated. + * Returns: + * NULL: Out of memory. + */ + struct ttm_tt *(*ttm_tt_create)(struct ttm_buffer_object *bo, + uint32_t page_flags); + + /** + * ttm_tt_populate + * + * @ttm: The struct ttm_tt to contain the backing pages. + * + * Allocate all backing pages + * Returns: + * -ENOMEM: Out of memory. + */ + int (*ttm_tt_populate)(struct ttm_device *bdev, + struct ttm_tt *ttm, + struct ttm_operation_ctx *ctx); + + /** + * ttm_tt_unpopulate + * + * @ttm: The struct ttm_tt to contain the backing pages. + * + * Free all backing page + */ + void (*ttm_tt_unpopulate)(struct ttm_device *bdev, + struct ttm_tt *ttm); + + /** + * ttm_tt_destroy + * + * @bdev: Pointer to a ttm device + * @ttm: Pointer to a struct ttm_tt. + * + * Destroy the backend. This will be call back from ttm_tt_destroy so + * don't call ttm_tt_destroy from the callback or infinite loop. + */ + void (*ttm_tt_destroy)(struct ttm_device *bdev, struct ttm_tt *ttm); + + /** + * struct ttm_bo_driver member eviction_valuable + * + * @bo: the buffer object to be evicted + * @place: placement we need room for + * + * Check with the driver if it is valuable to evict a BO to make room + * for a certain placement. + */ + bool (*eviction_valuable)(struct ttm_buffer_object *bo, + const struct ttm_place *place); + /** + * struct ttm_bo_driver member evict_flags: + * + * @bo: the buffer object to be evicted + * + * Return the bo flags for a buffer which is not mapped to the hardware. + * These will be placed in proposed_flags so that when the move is + * finished, they'll end up in bo->mem.flags + * This should not cause multihop evictions, and the core will warn + * if one is proposed. + */ + + void (*evict_flags)(struct ttm_buffer_object *bo, + struct ttm_placement *placement); + + /** + * struct ttm_bo_driver member move: + * + * @bo: the buffer to move + * @evict: whether this motion is evicting the buffer from + * the graphics address space + * @ctx: context for this move with parameters + * @new_mem: the new memory region receiving the buffer + @ @hop: placement for driver directed intermediate hop + * + * Move a buffer between two memory regions. + * Returns errno -EMULTIHOP if driver requests a hop + */ + int (*move)(struct ttm_buffer_object *bo, bool evict, + struct ttm_operation_ctx *ctx, + struct ttm_resource *new_mem, + struct ttm_place *hop); + + /** + * struct ttm_bo_driver_member verify_access + * + * @bo: Pointer to a buffer object. + * @filp: Pointer to a struct file trying to access the object. + * + * Called from the map / write / read methods to verify that the + * caller is permitted to access the buffer object. + * This member may be set to NULL, which will refuse this kind of + * access for all buffer objects. + * This function should return 0 if access is granted, -EPERM otherwise. + */ + int (*verify_access)(struct ttm_buffer_object *bo, + struct file *filp); + + /** + * Hook to notify driver about a resource delete. + */ + void (*delete_mem_notify)(struct ttm_buffer_object *bo); + + /** + * notify the driver that we're about to swap out this bo + */ + void (*swap_notify)(struct ttm_buffer_object *bo); + + /** + * Driver callback on when mapping io memory (for bo_move_memcpy + * for instance). TTM will take care to call io_mem_free whenever + * the mapping is not use anymore. io_mem_reserve & io_mem_free + * are balanced. + */ + int (*io_mem_reserve)(struct ttm_device *bdev, + struct ttm_resource *mem); + void (*io_mem_free)(struct ttm_device *bdev, + struct ttm_resource *mem); + + /** + * Return the pfn for a given page_offset inside the BO. + * + * @bo: the BO to look up the pfn for + * @page_offset: the offset to look up + */ + unsigned long (*io_mem_pfn)(struct ttm_buffer_object *bo, + unsigned long page_offset); + + /** + * Read/write memory buffers for ptrace access + * + * @bo: the BO to access + * @offset: the offset from the start of the BO + * @buf: pointer to source/destination buffer + * @len: number of bytes to copy + * @write: whether to read (0) from or write (non-0) to BO + * + * If successful, this function should return the number of + * bytes copied, -EIO otherwise. If the number of bytes + * returned is < len, the function may be called again with + * the remainder of the buffer to copy. + */ + int (*access_memory)(struct ttm_buffer_object *bo, unsigned long offset, + void *buf, int len, int write); + + /** + * struct ttm_bo_driver member del_from_lru_notify + * + * @bo: the buffer object deleted from lru + * + * notify driver that a BO was deleted from LRU. + */ + void (*del_from_lru_notify)(struct ttm_buffer_object *bo); + + /** + * Notify the driver that we're about to release a BO + * + * @bo: BO that is about to be released + * + * Gives the driver a chance to do any cleanup, including + * adding fences that may force a delayed delete + */ + void (*release_notify)(struct ttm_buffer_object *bo); +}; + +/** + * struct ttm_device - Buffer object driver device-specific data. + * + * @device_list: Our entry in the global device list. + * @funcs: Function table for the device. + * @sysman: Resource manager for the system domain. + * @man_drv: An array of resource_managers. + * @vma_manager: Address space manager. + * @pool: page pool for the device. + * @dev_mapping: A pointer to the struct address_space representing the + * device address space. + * @wq: Work queue structure for the delayed delete workqueue. + */ +struct ttm_device { + /* + * Constant after bo device init + */ + struct list_head device_list; + struct ttm_device_funcs *funcs; + + /* + * Access via ttm_manager_type. + */ + struct ttm_resource_manager sysman; + struct ttm_resource_manager *man_drv[TTM_NUM_MEM_TYPES]; + + /* + * Protected by internal locks. + */ + struct drm_vma_offset_manager *vma_manager; + struct ttm_pool pool; + + /* + * Protected by the global:lru lock. + */ + struct list_head ddestroy; + + /* + * Protected by load / firstopen / lastclose /unload sync. + */ + struct address_space *dev_mapping; + + /* + * Internal protection. + */ + struct delayed_work wq; +}; + +static inline struct ttm_resource_manager * +ttm_manager_type(struct ttm_device *bdev, int mem_type) +{ + return bdev->man_drv[mem_type]; +} + +static inline void ttm_set_driver_manager(struct ttm_device *bdev, int type, + struct ttm_resource_manager *manager) +{ + bdev->man_drv[type] = manager; +} + +int ttm_device_init(struct ttm_device *bdev, struct ttm_device_funcs *funcs, + struct device *dev, struct address_space *mapping, + struct drm_vma_offset_manager *vma_manager, + bool use_dma_alloc, bool use_dma32); +void ttm_device_fini(struct ttm_device *bdev); + +#endif diff --git a/include/drm/ttm/ttm_resource.h b/include/drm/ttm/ttm_resource.h index da0ed7e8c915..6164ccf4f308 100644 --- a/include/drm/ttm/ttm_resource.h +++ b/include/drm/ttm/ttm_resource.h @@ -33,7 +33,7 @@ #define TTM_MAX_BO_PRIORITY 4U -struct ttm_bo_device; +struct ttm_device; struct ttm_resource_manager; struct ttm_resource; struct ttm_place; @@ -233,7 +233,7 @@ void ttm_resource_free(struct ttm_buffer_object *bo, struct ttm_resource *res); void ttm_resource_manager_init(struct ttm_resource_manager *man, unsigned long p_size); -int ttm_resource_manager_evict_all(struct ttm_bo_device *bdev, +int ttm_resource_manager_evict_all(struct ttm_device *bdev, struct ttm_resource_manager *man); void ttm_resource_manager_debug(struct ttm_resource_manager *man, diff --git a/include/drm/ttm/ttm_tt.h b/include/drm/ttm/ttm_tt.h index 6c8eb9a4de81..0020a0588985 100644 --- a/include/drm/ttm/ttm_tt.h +++ b/include/drm/ttm/ttm_tt.h @@ -118,14 +118,14 @@ void ttm_tt_fini(struct ttm_tt *ttm); * * Unbind, unpopulate and destroy common struct ttm_tt. */ -void ttm_tt_destroy(struct ttm_bo_device *bdev, struct ttm_tt *ttm); +void ttm_tt_destroy(struct ttm_device *bdev, struct ttm_tt *ttm); /** * ttm_tt_destroy_common: * * Called from driver to destroy common path. */ -void ttm_tt_destroy_common(struct ttm_bo_device *bdev, struct ttm_tt *ttm); +void ttm_tt_destroy_common(struct ttm_device *bdev, struct ttm_tt *ttm); /** * ttm_tt_swapin: @@ -135,7 +135,7 @@ void ttm_tt_destroy_common(struct ttm_bo_device *bdev, struct ttm_tt *ttm); * Swap in a previously swap out ttm_tt. */ int ttm_tt_swapin(struct ttm_tt *ttm); -int ttm_tt_swapout(struct ttm_bo_device *bdev, struct ttm_tt *ttm); +int ttm_tt_swapout(struct ttm_device *bdev, struct ttm_tt *ttm); /** * ttm_tt_populate - allocate pages for a ttm @@ -144,7 +144,7 @@ int ttm_tt_swapout(struct ttm_bo_device *bdev, struct ttm_tt *ttm); * * Calls the driver method to allocate pages for a ttm */ -int ttm_tt_populate(struct ttm_bo_device *bdev, struct ttm_tt *ttm, struct ttm_operation_ctx *ctx); +int ttm_tt_populate(struct ttm_device *bdev, struct ttm_tt *ttm, struct ttm_operation_ctx *ctx); /** * ttm_tt_unpopulate - free pages from a ttm @@ -153,7 +153,7 @@ int ttm_tt_populate(struct ttm_bo_device *bdev, struct ttm_tt *ttm, struct ttm_o * * Calls the driver method to free all pages from a ttm */ -void ttm_tt_unpopulate(struct ttm_bo_device *bdev, struct ttm_tt *ttm); +void ttm_tt_unpopulate(struct ttm_device *bdev, struct ttm_tt *ttm); #if IS_ENABLED(CONFIG_AGP) #include -- cgit v1.2.3 From f3ebd4e6b692ab7af464561410a1f05dfc850823 Mon Sep 17 00:00:00 2001 From: Dan Carpenter Date: Mon, 25 Jan 2021 11:45:37 +0300 Subject: drm/vmwgfx/vmwgfx_drv: Fix an error path in vmw_setup_pci_resources() The devm_memremap() function never returns NULL, it returns error pointers so the test needs to be fixed. Also we need to call pci_release_regions() to avoid a memory leak. Fixes: be4f77ac6884 ("drm/vmwgfx: Cleanup fifo mmio handling") Signed-off-by: Dan Carpenter Signed-off-by: Zack Rusin Link: https://patchwork.freedesktop.org/patch/msgid/YA6FMboLhnE3uSvb@mwanda --- drivers/gpu/drm/vmwgfx/vmwgfx_drv.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'drivers/gpu/drm/vmwgfx/vmwgfx_drv.c') diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c index b454d80c273e..22f5a9f8a44b 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c @@ -668,9 +668,10 @@ static int vmw_setup_pci_resources(struct vmw_private *dev, fifo_size, MEMREMAP_WB); - if (unlikely(dev->fifo_mem == NULL)) { + if (IS_ERR(dev->fifo_mem)) { DRM_ERROR("Failed mapping FIFO memory.\n"); - return -ENOMEM; + pci_release_regions(pdev); + return PTR_ERR(dev->fifo_mem); } /* -- cgit v1.2.3 From 75ec69c79ebcae0ad16baf2249e378f5b02bdc12 Mon Sep 17 00:00:00 2001 From: Zack Rusin Date: Thu, 28 Jan 2021 12:35:22 -0500 Subject: drm/vmwgfx: Fix some memory leaks on errors Dan noticed some issues with pci_release_region, this builds upon that and fixes some other lingering issues. As part of this we also can stop trying to manually free our managed device; there's no need for it, it will be cleaned up automatically for us. Fixes: 8772c0bb58bb ("drm/vmwgfx: Cleanup pci resource allocation") Signed-off-by: Zack Rusin Reviewed-by: Martin Krastev Reviewed-by: Roland Scheidegger Reported-by: kernel test robot Reported-by: Dan Carpenter Link: https://patchwork.freedesktop.org/patch/msgid/20210128173756.121525-1-zackr@vmware.com --- drivers/gpu/drm/vmwgfx/vmwgfx_drv.c | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-) (limited to 'drivers/gpu/drm/vmwgfx/vmwgfx_drv.c') diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c index 22f5a9f8a44b..792feb08398b 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c @@ -717,7 +717,7 @@ static int vmw_driver_load(struct vmw_private *dev_priv, u32 pci_id) return ret; ret = vmw_detect_version(dev_priv); if (ret) - return ret; + goto out_no_pci_or_version; mutex_init(&dev_priv->cmdbuf_mutex); mutex_init(&dev_priv->release_mutex); @@ -1014,7 +1014,6 @@ out_no_fman: if (dev_priv->capabilities & SVGA_CAP_IRQMASK) vmw_irq_uninstall(&dev_priv->drm); out_no_irq: - pci_release_regions(pdev); ttm_object_device_release(&dev_priv->tdev); out_err0: for (i = vmw_res_context; i < vmw_res_max; ++i) @@ -1022,7 +1021,8 @@ out_err0: if (dev_priv->ctx.staged_bindings) vmw_binding_state_free(dev_priv->ctx.staged_bindings); - kfree(dev_priv); +out_no_pci_or_version: + pci_release_regions(pdev); return ret; } @@ -1060,7 +1060,6 @@ static void vmw_driver_unload(struct drm_device *dev) vmw_fence_manager_takedown(dev_priv->fman); if (dev_priv->capabilities & SVGA_CAP_IRQMASK) vmw_irq_uninstall(&dev_priv->drm); - pci_release_regions(pdev); ttm_object_device_release(&dev_priv->tdev); if (dev_priv->ctx.staged_bindings) @@ -1069,7 +1068,7 @@ static void vmw_driver_unload(struct drm_device *dev) for (i = vmw_res_context; i < vmw_res_max; ++i) idr_destroy(&dev_priv->res_idr[i]); - kfree(dev_priv); + pci_release_regions(pdev); } static void vmw_postclose(struct drm_device *dev, -- cgit v1.2.3 From d4bd7776a7ac504510656c0053a55c0cfd1ebc96 Mon Sep 17 00:00:00 2001 From: Christian König Date: Mon, 16 Nov 2020 13:47:11 +0100 Subject: drm/ttm: rework ttm_tt page limit v4 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit TTM implements a rather extensive accounting of allocated memory. There are two reasons for this: 1. It tries to block userspace allocating a huge number of very small BOs without accounting for the kmalloced memory. 2. Make sure we don't over allocate and run into an OOM situation during swapout while trying to handle the memory shortage. This is only partially a good idea. First of all it is perfectly valid for an application to use all of system memory, limiting it to 50% is not really acceptable. What we need to take care of is that the application is held accountable for the memory it allocated. This is what control mechanisms like memcg and the normal Linux page accounting already do. Making sure that we don't run into an OOM situation while trying to cope with a memory shortage is still a good idea, but this is also not very well implemented since it means another opportunity of recursion from the driver back into TTM. So start to rework all of this by implementing a shrinker callback which allows for TT object to be swapped out if necessary. v2: Switch from limit to shrinker callback. v3: fix gfp mask handling, use atomic for swapable_pages, add debugfs v4: drop the extra gfp_mask checks Signed-off-by: Christian König Reviewed-by: Daniel Vetter Link: https://patchwork.freedesktop.org/patch/msgid/20210208133226.36955-1-christian.koenig@amd.com --- drivers/gpu/drm/ttm/ttm_bo.c | 4 +- drivers/gpu/drm/ttm/ttm_memory.c | 7 ++- drivers/gpu/drm/ttm/ttm_tt.c | 105 +++++++++++++++++++++++++++++++++--- drivers/gpu/drm/vmwgfx/vmwgfx_drv.c | 2 +- include/drm/ttm/ttm_bo_api.h | 2 +- include/drm/ttm/ttm_tt.h | 6 ++- 6 files changed, 111 insertions(+), 15 deletions(-) (limited to 'drivers/gpu/drm/vmwgfx/vmwgfx_drv.c') diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index 20256797f3a6..643befc1a6f2 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -1219,7 +1219,7 @@ EXPORT_SYMBOL(ttm_bo_wait); * A buffer object shrink method that tries to swap out the first * buffer object on the bo_global::swap_lru list. */ -int ttm_bo_swapout(struct ttm_operation_ctx *ctx) +int ttm_bo_swapout(struct ttm_operation_ctx *ctx, gfp_t gfp_flags) { struct ttm_global *glob = &ttm_glob; struct ttm_buffer_object *bo; @@ -1302,7 +1302,7 @@ int ttm_bo_swapout(struct ttm_operation_ctx *ctx) if (bo->bdev->funcs->swap_notify) bo->bdev->funcs->swap_notify(bo); - ret = ttm_tt_swapout(bo->bdev, bo->ttm); + ret = ttm_tt_swapout(bo->bdev, bo->ttm, gfp_flags); out: /** diff --git a/drivers/gpu/drm/ttm/ttm_memory.c b/drivers/gpu/drm/ttm/ttm_memory.c index a3bfbd9cea68..634a85c2dc4c 100644 --- a/drivers/gpu/drm/ttm/ttm_memory.c +++ b/drivers/gpu/drm/ttm/ttm_memory.c @@ -37,6 +37,7 @@ #include #include #include +#include #include "ttm_module.h" @@ -276,9 +277,9 @@ static void ttm_shrink(struct ttm_mem_global *glob, bool from_wq, while (ttm_zones_above_swap_target(glob, from_wq, extra)) { spin_unlock(&glob->lock); - ret = ttm_bo_swapout(ctx); + ret = ttm_bo_swapout(ctx, GFP_KERNEL); spin_lock(&glob->lock); - if (unlikely(ret != 0)) + if (unlikely(ret < 0)) break; } @@ -453,6 +454,7 @@ int ttm_mem_global_init(struct ttm_mem_global *glob) zone->name, (unsigned long long)zone->max_mem >> 10); } ttm_pool_mgr_init(glob->zone_kernel->max_mem/(2*PAGE_SIZE)); + ttm_tt_mgr_init(); return 0; out_no_zone: ttm_mem_global_release(glob); @@ -466,6 +468,7 @@ void ttm_mem_global_release(struct ttm_mem_global *glob) /* let the page allocator first stop the shrink work. */ ttm_pool_mgr_fini(); + ttm_tt_mgr_fini(); flush_workqueue(glob->swap_queue); destroy_workqueue(glob->swap_queue); diff --git a/drivers/gpu/drm/ttm/ttm_tt.c b/drivers/gpu/drm/ttm/ttm_tt.c index 7782d5393c7c..2f0833c98d2c 100644 --- a/drivers/gpu/drm/ttm/ttm_tt.c +++ b/drivers/gpu/drm/ttm/ttm_tt.c @@ -38,6 +38,11 @@ #include #include +#include "ttm_module.h" + +static struct shrinker mm_shrinker; +static atomic_long_t swapable_pages; + /* * Allocates a ttm structure for the given BO. */ @@ -223,32 +228,41 @@ out_err: return ret; } -int ttm_tt_swapout(struct ttm_device *bdev, struct ttm_tt *ttm) +/** + * ttm_tt_swapout - swap out tt object + * + * @bdev: TTM device structure. + * @ttm: The struct ttm_tt. + * @gfp_flags: Flags to use for memory allocation. + * + * Swapout a TT object to a shmem_file, return number of pages swapped out or + * negative error code. + */ +int ttm_tt_swapout(struct ttm_device *bdev, struct ttm_tt *ttm, + gfp_t gfp_flags) { + loff_t size = (loff_t)ttm->num_pages << PAGE_SHIFT; struct address_space *swap_space; struct file *swap_storage; struct page *from_page; struct page *to_page; - gfp_t gfp_mask; int i, ret; - swap_storage = shmem_file_setup("ttm swap", - ttm->num_pages << PAGE_SHIFT, - 0); + swap_storage = shmem_file_setup("ttm swap", size, 0); if (IS_ERR(swap_storage)) { pr_err("Failed allocating swap storage\n"); return PTR_ERR(swap_storage); } swap_space = swap_storage->f_mapping; - gfp_mask = mapping_gfp_mask(swap_space); + gfp_flags &= mapping_gfp_mask(swap_space); for (i = 0; i < ttm->num_pages; ++i) { from_page = ttm->pages[i]; if (unlikely(from_page == NULL)) continue; - to_page = shmem_read_mapping_page_gfp(swap_space, i, gfp_mask); + to_page = shmem_read_mapping_page_gfp(swap_space, i, gfp_flags); if (IS_ERR(to_page)) { ret = PTR_ERR(to_page); goto out_err; @@ -263,7 +277,7 @@ int ttm_tt_swapout(struct ttm_device *bdev, struct ttm_tt *ttm) ttm->swap_storage = swap_storage; ttm->page_flags |= TTM_PAGE_FLAG_SWAPPED; - return 0; + return ttm->num_pages; out_err: fput(swap_storage); @@ -280,6 +294,8 @@ static void ttm_tt_add_mapping(struct ttm_device *bdev, struct ttm_tt *ttm) for (i = 0; i < ttm->num_pages; ++i) ttm->pages[i]->mapping = bdev->dev_mapping; + + atomic_long_add(ttm->num_pages, &swapable_pages); } int ttm_tt_populate(struct ttm_device *bdev, @@ -326,6 +342,8 @@ static void ttm_tt_clear_mapping(struct ttm_tt *ttm) (*page)->mapping = NULL; (*page++)->index = 0; } + + atomic_long_sub(ttm->num_pages, &swapable_pages); } void ttm_tt_unpopulate(struct ttm_device *bdev, @@ -341,3 +359,74 @@ void ttm_tt_unpopulate(struct ttm_device *bdev, ttm_pool_free(&bdev->pool, ttm); ttm->page_flags &= ~TTM_PAGE_FLAG_PRIV_POPULATED; } + +/* As long as pages are available make sure to release at least one */ +static unsigned long ttm_tt_shrinker_scan(struct shrinker *shrink, + struct shrink_control *sc) +{ + struct ttm_operation_ctx ctx = { + .no_wait_gpu = false + }; + int ret; + + ret = ttm_bo_swapout(&ctx, GFP_NOFS); + return ret < 0 ? SHRINK_EMPTY : ret; +} + +/* Return the number of pages available or SHRINK_EMPTY if we have none */ +static unsigned long ttm_tt_shrinker_count(struct shrinker *shrink, + struct shrink_control *sc) +{ + unsigned long num_pages; + + num_pages = atomic_long_read(&swapable_pages); + return num_pages ? num_pages : SHRINK_EMPTY; +} + +#ifdef CONFIG_DEBUG_FS + +/* Test the shrinker functions and dump the result */ +static int ttm_tt_debugfs_shrink_show(struct seq_file *m, void *data) +{ + struct shrink_control sc = { .gfp_mask = GFP_KERNEL }; + + fs_reclaim_acquire(GFP_KERNEL); + seq_printf(m, "%lu/%lu\n", ttm_tt_shrinker_count(&mm_shrinker, &sc), + ttm_tt_shrinker_scan(&mm_shrinker, &sc)); + fs_reclaim_release(GFP_KERNEL); + + return 0; +} +DEFINE_SHOW_ATTRIBUTE(ttm_tt_debugfs_shrink); + +#endif + + + +/** + * ttm_tt_mgr_init - register with the MM shrinker + * + * Register with the MM shrinker for swapping out BOs. + */ +int ttm_tt_mgr_init(void) +{ +#ifdef CONFIG_DEBUG_FS + debugfs_create_file("tt_shrink", 0400, ttm_debugfs_root, NULL, + &ttm_tt_debugfs_shrink_fops); +#endif + + mm_shrinker.count_objects = ttm_tt_shrinker_count; + mm_shrinker.scan_objects = ttm_tt_shrinker_scan; + mm_shrinker.seeks = 1; + return register_shrinker(&mm_shrinker); +} + +/** + * ttm_tt_mgr_fini - unregister our MM shrinker + * + * Unregisters the MM shrinker. + */ +void ttm_tt_mgr_fini(void) +{ + unregister_shrinker(&mm_shrinker); +} diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c index 792feb08398b..9a9afa087802 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c @@ -1383,7 +1383,7 @@ static int vmw_pm_freeze(struct device *kdev) vmw_execbuf_release_pinned_bo(dev_priv); vmw_resource_evict_all(dev_priv); vmw_release_device_early(dev_priv); - while (ttm_bo_swapout(&ctx) == 0); + while (ttm_bo_swapout(&ctx, GFP_KERNEL) > 0); if (dev_priv->enable_fb) vmw_fifo_resource_dec(dev_priv); if (atomic_read(&dev_priv->num_fifo_resources) != 0) { diff --git a/include/drm/ttm/ttm_bo_api.h b/include/drm/ttm/ttm_bo_api.h index 62734db0b421..1297a8fb7ccb 100644 --- a/include/drm/ttm/ttm_bo_api.h +++ b/include/drm/ttm/ttm_bo_api.h @@ -569,7 +569,7 @@ ssize_t ttm_bo_io(struct ttm_device *bdev, struct file *filp, const char __user *wbuf, char __user *rbuf, size_t count, loff_t *f_pos, bool write); -int ttm_bo_swapout(struct ttm_operation_ctx *ctx); +int ttm_bo_swapout(struct ttm_operation_ctx *ctx, gfp_t gfp_flags); /** * ttm_bo_uses_embedded_gem_object - check if the given bo uses the diff --git a/include/drm/ttm/ttm_tt.h b/include/drm/ttm/ttm_tt.h index 0020a0588985..cce57fb49e2c 100644 --- a/include/drm/ttm/ttm_tt.h +++ b/include/drm/ttm/ttm_tt.h @@ -135,7 +135,8 @@ void ttm_tt_destroy_common(struct ttm_device *bdev, struct ttm_tt *ttm); * Swap in a previously swap out ttm_tt. */ int ttm_tt_swapin(struct ttm_tt *ttm); -int ttm_tt_swapout(struct ttm_device *bdev, struct ttm_tt *ttm); +int ttm_tt_swapout(struct ttm_device *bdev, struct ttm_tt *ttm, + gfp_t gfp_flags); /** * ttm_tt_populate - allocate pages for a ttm @@ -155,6 +156,9 @@ int ttm_tt_populate(struct ttm_device *bdev, struct ttm_tt *ttm, struct ttm_oper */ void ttm_tt_unpopulate(struct ttm_device *bdev, struct ttm_tt *ttm); +int ttm_tt_mgr_init(void); +void ttm_tt_mgr_fini(void); + #if IS_ENABLED(CONFIG_AGP) #include -- cgit v1.2.3 From f07069da6b4c5f937d6df2de6504394845513964 Mon Sep 17 00:00:00 2001 From: Christian König Date: Tue, 17 Nov 2020 13:52:28 +0100 Subject: drm/ttm: move memory accounting into vmwgfx v4 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit This is just another feature which is only used by VMWGFX, so move it into the driver instead. I've tried to add the accounting sysfs file to the kobject of the drm minor, but I'm not 100% sure if this works as expected. v2: fix typo in KFD and avoid 64bit divide v3: fix init order in VMWGFX v4: use pdev sysfs reference instead of drm Signed-off-by: Christian König Reviewed-by: Zack Rusin (v3) Tested-by: Nirmoy Das Link: https://patchwork.freedesktop.org/patch/msgid/20210208133226.36955-2-christian.koenig@amd.com --- drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c | 16 +- drivers/gpu/drm/amd/amdgpu/amdgpu_object.c | 8 +- drivers/gpu/drm/drm_gem_vram_helper.c | 6 +- drivers/gpu/drm/nouveau/nouveau_bo.c | 7 +- drivers/gpu/drm/nouveau/nouveau_drv.h | 1 - drivers/gpu/drm/qxl/qxl_object.c | 4 +- drivers/gpu/drm/radeon/radeon_object.c | 8 +- drivers/gpu/drm/ttm/Makefile | 7 +- drivers/gpu/drm/ttm/ttm_bo.c | 33 +- drivers/gpu/drm/ttm/ttm_bo_util.c | 1 - drivers/gpu/drm/ttm/ttm_device.c | 22 +- drivers/gpu/drm/ttm/ttm_memory.c | 686 ----------------------- drivers/gpu/drm/ttm/ttm_pool.c | 13 +- drivers/gpu/drm/vmwgfx/Makefile | 2 +- drivers/gpu/drm/vmwgfx/ttm_memory.c | 681 ++++++++++++++++++++++ drivers/gpu/drm/vmwgfx/ttm_memory.h | 96 ++++ drivers/gpu/drm/vmwgfx/ttm_object.h | 3 +- drivers/gpu/drm/vmwgfx/vmwgfx_bo.c | 22 +- drivers/gpu/drm/vmwgfx/vmwgfx_drv.c | 5 + drivers/gpu/drm/vmwgfx/vmwgfx_ttm_buffer.c | 28 +- include/drm/ttm/ttm_bo_api.h | 13 +- include/drm/ttm/ttm_bo_driver.h | 1 - include/drm/ttm/ttm_memory.h | 95 ---- include/drm/ttm/ttm_tt.h | 1 + 24 files changed, 877 insertions(+), 882 deletions(-) delete mode 100644 drivers/gpu/drm/ttm/ttm_memory.c create mode 100644 drivers/gpu/drm/vmwgfx/ttm_memory.c create mode 100644 drivers/gpu/drm/vmwgfx/ttm_memory.h delete mode 100644 include/drm/ttm/ttm_memory.h (limited to 'drivers/gpu/drm/vmwgfx/vmwgfx_drv.c') diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c index 0849b68e784f..e440af37dde8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c @@ -118,6 +118,16 @@ void amdgpu_amdkfd_gpuvm_init_mem_limits(void) */ #define ESTIMATE_PT_SIZE(mem_size) ((mem_size) >> 14) +static size_t amdgpu_amdkfd_acc_size(uint64_t size) +{ + size >>= PAGE_SHIFT; + size *= sizeof(dma_addr_t) + sizeof(void *); + + return __roundup_pow_of_two(sizeof(struct amdgpu_bo)) + + __roundup_pow_of_two(sizeof(struct ttm_tt)) + + PAGE_ALIGN(size); +} + static int amdgpu_amdkfd_reserve_mem_limit(struct amdgpu_device *adev, uint64_t size, u32 domain, bool sg) { @@ -126,8 +136,7 @@ static int amdgpu_amdkfd_reserve_mem_limit(struct amdgpu_device *adev, size_t acc_size, system_mem_needed, ttm_mem_needed, vram_needed; int ret = 0; - acc_size = ttm_bo_dma_acc_size(&adev->mman.bdev, size, - sizeof(struct amdgpu_bo)); + acc_size = amdgpu_amdkfd_acc_size(size); vram_needed = 0; if (domain == AMDGPU_GEM_DOMAIN_GTT) { @@ -174,8 +183,7 @@ static void unreserve_mem_limit(struct amdgpu_device *adev, { size_t acc_size; - acc_size = ttm_bo_dma_acc_size(&adev->mman.bdev, size, - sizeof(struct amdgpu_bo)); + acc_size = amdgpu_amdkfd_acc_size(size); spin_lock(&kfd_mem_limit.mem_limit_lock); if (domain == AMDGPU_GEM_DOMAIN_GTT) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c index 6cc9919b12cc..599c9a132eb6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c @@ -523,7 +523,6 @@ static int amdgpu_bo_do_create(struct amdgpu_device *adev, }; struct amdgpu_bo *bo; unsigned long page_align, size = bp->size; - size_t acc_size; int r; /* Note that GDS/GWS/OA allocates 1 page per byte/resource. */ @@ -546,9 +545,6 @@ static int amdgpu_bo_do_create(struct amdgpu_device *adev, *bo_ptr = NULL; - acc_size = ttm_bo_dma_acc_size(&adev->mman.bdev, size, - sizeof(struct amdgpu_bo)); - bo = kzalloc(sizeof(struct amdgpu_bo), GFP_KERNEL); if (bo == NULL) return -ENOMEM; @@ -577,8 +573,8 @@ static int amdgpu_bo_do_create(struct amdgpu_device *adev, bo->tbo.priority = 1; r = ttm_bo_init_reserved(&adev->mman.bdev, &bo->tbo, size, bp->type, - &bo->placement, page_align, &ctx, acc_size, - NULL, bp->resv, &amdgpu_bo_destroy); + &bo->placement, page_align, &ctx, NULL, + bp->resv, &amdgpu_bo_destroy); if (unlikely(r != 0)) return r; diff --git a/drivers/gpu/drm/drm_gem_vram_helper.c b/drivers/gpu/drm/drm_gem_vram_helper.c index 0b13c8507688..a0992f0b8afd 100644 --- a/drivers/gpu/drm/drm_gem_vram_helper.c +++ b/drivers/gpu/drm/drm_gem_vram_helper.c @@ -189,7 +189,6 @@ struct drm_gem_vram_object *drm_gem_vram_create(struct drm_device *dev, struct drm_vram_mm *vmm = dev->vram_mm; struct ttm_device *bdev; int ret; - size_t acc_size; if (WARN_ONCE(!vmm, "VRAM MM not initialized")) return ERR_PTR(-EINVAL); @@ -216,7 +215,6 @@ struct drm_gem_vram_object *drm_gem_vram_create(struct drm_device *dev, } bdev = &vmm->bdev; - acc_size = ttm_bo_dma_acc_size(bdev, size, sizeof(*gbo)); gbo->bo.bdev = bdev; drm_gem_vram_placement(gbo, DRM_GEM_VRAM_PL_FLAG_SYSTEM); @@ -226,8 +224,8 @@ struct drm_gem_vram_object *drm_gem_vram_create(struct drm_device *dev, * to release gbo->bo.base and kfree gbo. */ ret = ttm_bo_init(bdev, &gbo->bo, size, ttm_bo_type_device, - &gbo->placement, pg_align, false, acc_size, - NULL, NULL, ttm_buffer_object_destroy); + &gbo->placement, pg_align, false, NULL, NULL, + ttm_buffer_object_destroy); if (ret) return ERR_PTR(ret); diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.c b/drivers/gpu/drm/nouveau/nouveau_bo.c index c177940d6e2c..ca2a8ae1938e 100644 --- a/drivers/gpu/drm/nouveau/nouveau_bo.c +++ b/drivers/gpu/drm/nouveau/nouveau_bo.c @@ -300,18 +300,15 @@ nouveau_bo_init(struct nouveau_bo *nvbo, u64 size, int align, u32 domain, struct sg_table *sg, struct dma_resv *robj) { int type = sg ? ttm_bo_type_sg : ttm_bo_type_device; - size_t acc_size; int ret; - acc_size = ttm_bo_dma_acc_size(nvbo->bo.bdev, size, sizeof(*nvbo)); - nvbo->bo.mem.num_pages = size >> PAGE_SHIFT; nouveau_bo_placement_set(nvbo, domain, 0); INIT_LIST_HEAD(&nvbo->io_reserve_lru); ret = ttm_bo_init(nvbo->bo.bdev, &nvbo->bo, size, type, - &nvbo->placement, align >> PAGE_SHIFT, false, - acc_size, sg, robj, nouveau_bo_del_ttm); + &nvbo->placement, align >> PAGE_SHIFT, false, sg, + robj, nouveau_bo_del_ttm); if (ret) { /* ttm will call nouveau_bo_del_ttm if it fails.. */ return ret; diff --git a/drivers/gpu/drm/nouveau/nouveau_drv.h b/drivers/gpu/drm/nouveau/nouveau_drv.h index edf9d1ee9d58..a491c2c1c56e 100644 --- a/drivers/gpu/drm/nouveau/nouveau_drv.h +++ b/drivers/gpu/drm/nouveau/nouveau_drv.h @@ -54,7 +54,6 @@ #include #include #include -#include #include diff --git a/drivers/gpu/drm/qxl/qxl_object.c b/drivers/gpu/drm/qxl/qxl_object.c index ceebc5881f68..705b51535492 100644 --- a/drivers/gpu/drm/qxl/qxl_object.c +++ b/drivers/gpu/drm/qxl/qxl_object.c @@ -138,8 +138,8 @@ int qxl_bo_create(struct qxl_device *qdev, qxl_ttm_placement_from_domain(bo, domain); r = ttm_bo_init_reserved(&qdev->mman.bdev, &bo->tbo, size, type, - &bo->placement, 0, &ctx, size, - NULL, NULL, &qxl_ttm_bo_destroy); + &bo->placement, 0, &ctx, NULL, NULL, + &qxl_ttm_bo_destroy); if (unlikely(r != 0)) { if (r != -ERESTARTSYS) dev_err(qdev->ddev.dev, diff --git a/drivers/gpu/drm/radeon/radeon_object.c b/drivers/gpu/drm/radeon/radeon_object.c index 6a336284466f..804f7a427be7 100644 --- a/drivers/gpu/drm/radeon/radeon_object.c +++ b/drivers/gpu/drm/radeon/radeon_object.c @@ -159,7 +159,6 @@ int radeon_bo_create(struct radeon_device *rdev, struct radeon_bo *bo; enum ttm_bo_type type; unsigned long page_align = roundup(byte_align, PAGE_SIZE) >> PAGE_SHIFT; - size_t acc_size; int r; size = ALIGN(size, PAGE_SIZE); @@ -173,9 +172,6 @@ int radeon_bo_create(struct radeon_device *rdev, } *bo_ptr = NULL; - acc_size = ttm_bo_dma_acc_size(&rdev->mman.bdev, size, - sizeof(struct radeon_bo)); - bo = kzalloc(sizeof(struct radeon_bo), GFP_KERNEL); if (bo == NULL) return -ENOMEM; @@ -230,8 +226,8 @@ int radeon_bo_create(struct radeon_device *rdev, /* Kernel allocation are uninterruptible */ down_read(&rdev->pm.mclk_lock); r = ttm_bo_init(&rdev->mman.bdev, &bo->tbo, size, type, - &bo->placement, page_align, !kernel, acc_size, - sg, resv, &radeon_ttm_bo_destroy); + &bo->placement, page_align, !kernel, sg, resv, + &radeon_ttm_bo_destroy); up_read(&rdev->pm.mclk_lock); if (unlikely(r != 0)) { return r; diff --git a/drivers/gpu/drm/ttm/Makefile b/drivers/gpu/drm/ttm/Makefile index 8e6437eadabe..40e5e9da7953 100644 --- a/drivers/gpu/drm/ttm/Makefile +++ b/drivers/gpu/drm/ttm/Makefile @@ -2,10 +2,9 @@ # # Makefile for the drm device driver. This driver provides support for the -ttm-y := ttm_memory.o ttm_tt.o ttm_bo.o \ - ttm_bo_util.o ttm_bo_vm.o ttm_module.o \ - ttm_execbuf_util.o ttm_range_manager.o \ - ttm_resource.o ttm_pool.o ttm_device.o +ttm-y := ttm_tt.o ttm_bo.o ttm_bo_util.o ttm_bo_vm.o ttm_module.o \ + ttm_execbuf_util.o ttm_range_manager.o ttm_resource.o ttm_pool.o \ + ttm_device.o ttm-$(CONFIG_AGP) += ttm_agp_backend.o obj-$(CONFIG_DRM_TTM) += ttm.o diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index 643befc1a6f2..e38102282fd5 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -425,7 +425,6 @@ static void ttm_bo_release(struct kref *kref) struct ttm_buffer_object *bo = container_of(kref, struct ttm_buffer_object, kref); struct ttm_device *bdev = bo->bdev; - size_t acc_size = bo->acc_size; int ret; if (!bo->deleted) { @@ -485,7 +484,6 @@ static void ttm_bo_release(struct kref *kref) if (!ttm_bo_uses_embedded_gem_object(bo)) dma_resv_fini(&bo->base._resv); bo->destroy(bo); - ttm_mem_global_free(&ttm_mem_glob, acc_size); } void ttm_bo_put(struct ttm_buffer_object *bo) @@ -1046,25 +1044,13 @@ int ttm_bo_init_reserved(struct ttm_device *bdev, struct ttm_placement *placement, uint32_t page_alignment, struct ttm_operation_ctx *ctx, - size_t acc_size, struct sg_table *sg, struct dma_resv *resv, void (*destroy) (struct ttm_buffer_object *)) { - struct ttm_mem_global *mem_glob = &ttm_mem_glob; bool locked; int ret = 0; - ret = ttm_mem_global_alloc(mem_glob, acc_size, ctx); - if (ret) { - pr_err("Out of kernel memory\n"); - if (destroy) - (*destroy)(bo); - else - kfree(bo); - return -ENOMEM; - } - bo->destroy = destroy ? destroy : ttm_bo_default_destroy; kref_init(&bo->kref); @@ -1081,7 +1067,6 @@ int ttm_bo_init_reserved(struct ttm_device *bdev, bo->mem.bus.addr = NULL; bo->moving = NULL; bo->mem.placement = 0; - bo->acc_size = acc_size; bo->pin_count = 0; bo->sg = sg; if (resv) { @@ -1142,7 +1127,6 @@ int ttm_bo_init(struct ttm_device *bdev, struct ttm_placement *placement, uint32_t page_alignment, bool interruptible, - size_t acc_size, struct sg_table *sg, struct dma_resv *resv, void (*destroy) (struct ttm_buffer_object *)) @@ -1151,8 +1135,7 @@ int ttm_bo_init(struct ttm_device *bdev, int ret; ret = ttm_bo_init_reserved(bdev, bo, size, type, placement, - page_alignment, &ctx, acc_size, - sg, resv, destroy); + page_alignment, &ctx, sg, resv, destroy); if (ret) return ret; @@ -1163,20 +1146,6 @@ int ttm_bo_init(struct ttm_device *bdev, } EXPORT_SYMBOL(ttm_bo_init); -size_t ttm_bo_dma_acc_size(struct ttm_device *bdev, - unsigned long bo_size, - unsigned struct_size) -{ - unsigned npages = (PAGE_ALIGN(bo_size)) >> PAGE_SHIFT; - size_t size = 0; - - size += ttm_round_pot(struct_size); - size += ttm_round_pot(npages * (2*sizeof(void *) + sizeof(dma_addr_t))); - size += ttm_round_pot(sizeof(struct ttm_tt)); - return size; -} -EXPORT_SYMBOL(ttm_bo_dma_acc_size); - /* * buffer object vm functions. */ diff --git a/drivers/gpu/drm/ttm/ttm_bo_util.c b/drivers/gpu/drm/ttm/ttm_bo_util.c index db0f2661d504..031e5819fec4 100644 --- a/drivers/gpu/drm/ttm/ttm_bo_util.c +++ b/drivers/gpu/drm/ttm/ttm_bo_util.c @@ -309,7 +309,6 @@ static int ttm_buffer_object_transfer(struct ttm_buffer_object *bo, kref_init(&fbo->base.kref); fbo->base.destroy = &ttm_transfered_destroy; - fbo->base.acc_size = 0; fbo->base.pin_count = 0; if (bo->type != ttm_bo_type_sg) fbo->base.base.resv = &fbo->base.base._resv; diff --git a/drivers/gpu/drm/ttm/ttm_device.c b/drivers/gpu/drm/ttm/ttm_device.c index 8bb61dd26437..95e1b7b1f2e6 100644 --- a/drivers/gpu/drm/ttm/ttm_device.c +++ b/drivers/gpu/drm/ttm/ttm_device.c @@ -27,9 +27,12 @@ #define pr_fmt(fmt) "[TTM DEVICE] " fmt +#include + #include -#include +#include #include +#include #include "ttm_module.h" @@ -49,7 +52,9 @@ static void ttm_global_release(void) if (--ttm_glob_use_count > 0) goto out; - ttm_mem_global_release(&ttm_mem_glob); + ttm_pool_mgr_fini(); + ttm_tt_mgr_fini(); + __free_page(glob->dummy_read_page); memset(glob, 0, sizeof(*glob)); out: @@ -59,6 +64,8 @@ out: static int ttm_global_init(void) { struct ttm_global *glob = &ttm_glob; + unsigned long num_pages; + struct sysinfo si; int ret = 0; unsigned i; @@ -66,9 +73,14 @@ static int ttm_global_init(void) if (++ttm_glob_use_count > 1) goto out; - ret = ttm_mem_global_init(&ttm_mem_glob); - if (ret) - goto out; + si_meminfo(&si); + + /* Limit the number of pages in the pool to about 50% of the total + * system memory. + */ + num_pages = ((u64)si.totalram * si.mem_unit) >> PAGE_SHIFT; + ttm_pool_mgr_init(num_pages * 50 / 100); + ttm_tt_mgr_init(); spin_lock_init(&glob->lru_lock); glob->dummy_read_page = alloc_page(__GFP_ZERO | GFP_DMA32); diff --git a/drivers/gpu/drm/ttm/ttm_memory.c b/drivers/gpu/drm/ttm/ttm_memory.c deleted file mode 100644 index 634a85c2dc4c..000000000000 --- a/drivers/gpu/drm/ttm/ttm_memory.c +++ /dev/null @@ -1,686 +0,0 @@ -/* SPDX-License-Identifier: GPL-2.0 OR MIT */ -/************************************************************************** - * - * Copyright (c) 2006-2009 VMware, Inc., Palo Alto, CA., USA - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, - * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR - * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE - * USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - -#define pr_fmt(fmt) "[TTM] " fmt - -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include - -#include "ttm_module.h" - -#define TTM_MEMORY_ALLOC_RETRIES 4 - -struct ttm_mem_global ttm_mem_glob; -EXPORT_SYMBOL(ttm_mem_glob); - -struct ttm_mem_zone { - struct kobject kobj; - struct ttm_mem_global *glob; - const char *name; - uint64_t zone_mem; - uint64_t emer_mem; - uint64_t max_mem; - uint64_t swap_limit; - uint64_t used_mem; -}; - -static struct attribute ttm_mem_sys = { - .name = "zone_memory", - .mode = S_IRUGO -}; -static struct attribute ttm_mem_emer = { - .name = "emergency_memory", - .mode = S_IRUGO | S_IWUSR -}; -static struct attribute ttm_mem_max = { - .name = "available_memory", - .mode = S_IRUGO | S_IWUSR -}; -static struct attribute ttm_mem_swap = { - .name = "swap_limit", - .mode = S_IRUGO | S_IWUSR -}; -static struct attribute ttm_mem_used = { - .name = "used_memory", - .mode = S_IRUGO -}; - -static void ttm_mem_zone_kobj_release(struct kobject *kobj) -{ - struct ttm_mem_zone *zone = - container_of(kobj, struct ttm_mem_zone, kobj); - - pr_info("Zone %7s: Used memory at exit: %llu KiB\n", - zone->name, (unsigned long long)zone->used_mem >> 10); - kfree(zone); -} - -static ssize_t ttm_mem_zone_show(struct kobject *kobj, - struct attribute *attr, - char *buffer) -{ - struct ttm_mem_zone *zone = - container_of(kobj, struct ttm_mem_zone, kobj); - uint64_t val = 0; - - spin_lock(&zone->glob->lock); - if (attr == &ttm_mem_sys) - val = zone->zone_mem; - else if (attr == &ttm_mem_emer) - val = zone->emer_mem; - else if (attr == &ttm_mem_max) - val = zone->max_mem; - else if (attr == &ttm_mem_swap) - val = zone->swap_limit; - else if (attr == &ttm_mem_used) - val = zone->used_mem; - spin_unlock(&zone->glob->lock); - - return snprintf(buffer, PAGE_SIZE, "%llu\n", - (unsigned long long) val >> 10); -} - -static void ttm_check_swapping(struct ttm_mem_global *glob); - -static ssize_t ttm_mem_zone_store(struct kobject *kobj, - struct attribute *attr, - const char *buffer, - size_t size) -{ - struct ttm_mem_zone *zone = - container_of(kobj, struct ttm_mem_zone, kobj); - int chars; - unsigned long val; - uint64_t val64; - - chars = sscanf(buffer, "%lu", &val); - if (chars == 0) - return size; - - val64 = val; - val64 <<= 10; - - spin_lock(&zone->glob->lock); - if (val64 > zone->zone_mem) - val64 = zone->zone_mem; - if (attr == &ttm_mem_emer) { - zone->emer_mem = val64; - if (zone->max_mem > val64) - zone->max_mem = val64; - } else if (attr == &ttm_mem_max) { - zone->max_mem = val64; - if (zone->emer_mem < val64) - zone->emer_mem = val64; - } else if (attr == &ttm_mem_swap) - zone->swap_limit = val64; - spin_unlock(&zone->glob->lock); - - ttm_check_swapping(zone->glob); - - return size; -} - -static struct attribute *ttm_mem_zone_attrs[] = { - &ttm_mem_sys, - &ttm_mem_emer, - &ttm_mem_max, - &ttm_mem_swap, - &ttm_mem_used, - NULL -}; - -static const struct sysfs_ops ttm_mem_zone_ops = { - .show = &ttm_mem_zone_show, - .store = &ttm_mem_zone_store -}; - -static struct kobj_type ttm_mem_zone_kobj_type = { - .release = &ttm_mem_zone_kobj_release, - .sysfs_ops = &ttm_mem_zone_ops, - .default_attrs = ttm_mem_zone_attrs, -}; - -static struct attribute ttm_mem_global_lower_mem_limit = { - .name = "lower_mem_limit", - .mode = S_IRUGO | S_IWUSR -}; - -static ssize_t ttm_mem_global_show(struct kobject *kobj, - struct attribute *attr, - char *buffer) -{ - struct ttm_mem_global *glob = - container_of(kobj, struct ttm_mem_global, kobj); - uint64_t val = 0; - - spin_lock(&glob->lock); - val = glob->lower_mem_limit; - spin_unlock(&glob->lock); - /* convert from number of pages to KB */ - val <<= (PAGE_SHIFT - 10); - return snprintf(buffer, PAGE_SIZE, "%llu\n", - (unsigned long long) val); -} - -static ssize_t ttm_mem_global_store(struct kobject *kobj, - struct attribute *attr, - const char *buffer, - size_t size) -{ - int chars; - uint64_t val64; - unsigned long val; - struct ttm_mem_global *glob = - container_of(kobj, struct ttm_mem_global, kobj); - - chars = sscanf(buffer, "%lu", &val); - if (chars == 0) - return size; - - val64 = val; - /* convert from KB to number of pages */ - val64 >>= (PAGE_SHIFT - 10); - - spin_lock(&glob->lock); - glob->lower_mem_limit = val64; - spin_unlock(&glob->lock); - - return size; -} - -static struct attribute *ttm_mem_global_attrs[] = { - &ttm_mem_global_lower_mem_limit, - NULL -}; - -static const struct sysfs_ops ttm_mem_global_ops = { - .show = &ttm_mem_global_show, - .store = &ttm_mem_global_store, -}; - -static struct kobj_type ttm_mem_glob_kobj_type = { - .sysfs_ops = &ttm_mem_global_ops, - .default_attrs = ttm_mem_global_attrs, -}; - -static bool ttm_zones_above_swap_target(struct ttm_mem_global *glob, - bool from_wq, uint64_t extra) -{ - unsigned int i; - struct ttm_mem_zone *zone; - uint64_t target; - - for (i = 0; i < glob->num_zones; ++i) { - zone = glob->zones[i]; - - if (from_wq) - target = zone->swap_limit; - else if (capable(CAP_SYS_ADMIN)) - target = zone->emer_mem; - else - target = zone->max_mem; - - target = (extra > target) ? 0ULL : target; - - if (zone->used_mem > target) - return true; - } - return false; -} - -/* - * At this point we only support a single shrink callback. - * Extend this if needed, perhaps using a linked list of callbacks. - * Note that this function is reentrant: - * many threads may try to swap out at any given time. - */ - -static void ttm_shrink(struct ttm_mem_global *glob, bool from_wq, - uint64_t extra, struct ttm_operation_ctx *ctx) -{ - int ret; - - spin_lock(&glob->lock); - - while (ttm_zones_above_swap_target(glob, from_wq, extra)) { - spin_unlock(&glob->lock); - ret = ttm_bo_swapout(ctx, GFP_KERNEL); - spin_lock(&glob->lock); - if (unlikely(ret < 0)) - break; - } - - spin_unlock(&glob->lock); -} - -static void ttm_shrink_work(struct work_struct *work) -{ - struct ttm_operation_ctx ctx = { - .interruptible = false, - .no_wait_gpu = false - }; - struct ttm_mem_global *glob = - container_of(work, struct ttm_mem_global, work); - - ttm_shrink(glob, true, 0ULL, &ctx); -} - -static int ttm_mem_init_kernel_zone(struct ttm_mem_global *glob, - const struct sysinfo *si) -{ - struct ttm_mem_zone *zone = kzalloc(sizeof(*zone), GFP_KERNEL); - uint64_t mem; - int ret; - - if (unlikely(!zone)) - return -ENOMEM; - - mem = si->totalram - si->totalhigh; - mem *= si->mem_unit; - - zone->name = "kernel"; - zone->zone_mem = mem; - zone->max_mem = mem >> 1; - zone->emer_mem = (mem >> 1) + (mem >> 2); - zone->swap_limit = zone->max_mem - (mem >> 3); - zone->used_mem = 0; - zone->glob = glob; - glob->zone_kernel = zone; - ret = kobject_init_and_add( - &zone->kobj, &ttm_mem_zone_kobj_type, &glob->kobj, zone->name); - if (unlikely(ret != 0)) { - kobject_put(&zone->kobj); - return ret; - } - glob->zones[glob->num_zones++] = zone; - return 0; -} - -#ifdef CONFIG_HIGHMEM -static int ttm_mem_init_highmem_zone(struct ttm_mem_global *glob, - const struct sysinfo *si) -{ - struct ttm_mem_zone *zone; - uint64_t mem; - int ret; - - if (si->totalhigh == 0) - return 0; - - zone = kzalloc(sizeof(*zone), GFP_KERNEL); - if (unlikely(!zone)) - return -ENOMEM; - - mem = si->totalram; - mem *= si->mem_unit; - - zone->name = "highmem"; - zone->zone_mem = mem; - zone->max_mem = mem >> 1; - zone->emer_mem = (mem >> 1) + (mem >> 2); - zone->swap_limit = zone->max_mem - (mem >> 3); - zone->used_mem = 0; - zone->glob = glob; - glob->zone_highmem = zone; - ret = kobject_init_and_add( - &zone->kobj, &ttm_mem_zone_kobj_type, &glob->kobj, "%s", - zone->name); - if (unlikely(ret != 0)) { - kobject_put(&zone->kobj); - return ret; - } - glob->zones[glob->num_zones++] = zone; - return 0; -} -#else -static int ttm_mem_init_dma32_zone(struct ttm_mem_global *glob, - const struct sysinfo *si) -{ - struct ttm_mem_zone *zone = kzalloc(sizeof(*zone), GFP_KERNEL); - uint64_t mem; - int ret; - - if (unlikely(!zone)) - return -ENOMEM; - - mem = si->totalram; - mem *= si->mem_unit; - - /** - * No special dma32 zone needed. - */ - - if (mem <= ((uint64_t) 1ULL << 32)) { - kfree(zone); - return 0; - } - - /* - * Limit max dma32 memory to 4GB for now - * until we can figure out how big this - * zone really is. - */ - - mem = ((uint64_t) 1ULL << 32); - zone->name = "dma32"; - zone->zone_mem = mem; - zone->max_mem = mem >> 1; - zone->emer_mem = (mem >> 1) + (mem >> 2); - zone->swap_limit = zone->max_mem - (mem >> 3); - zone->used_mem = 0; - zone->glob = glob; - glob->zone_dma32 = zone; - ret = kobject_init_and_add( - &zone->kobj, &ttm_mem_zone_kobj_type, &glob->kobj, zone->name); - if (unlikely(ret != 0)) { - kobject_put(&zone->kobj); - return ret; - } - glob->zones[glob->num_zones++] = zone; - return 0; -} -#endif - -int ttm_mem_global_init(struct ttm_mem_global *glob) -{ - struct sysinfo si; - int ret; - int i; - struct ttm_mem_zone *zone; - - spin_lock_init(&glob->lock); - glob->swap_queue = create_singlethread_workqueue("ttm_swap"); - INIT_WORK(&glob->work, ttm_shrink_work); - ret = kobject_init_and_add( - &glob->kobj, &ttm_mem_glob_kobj_type, ttm_get_kobj(), "memory_accounting"); - if (unlikely(ret != 0)) { - kobject_put(&glob->kobj); - return ret; - } - - si_meminfo(&si); - - /* set it as 0 by default to keep original behavior of OOM */ - glob->lower_mem_limit = 0; - - ret = ttm_mem_init_kernel_zone(glob, &si); - if (unlikely(ret != 0)) - goto out_no_zone; -#ifdef CONFIG_HIGHMEM - ret = ttm_mem_init_highmem_zone(glob, &si); - if (unlikely(ret != 0)) - goto out_no_zone; -#else - ret = ttm_mem_init_dma32_zone(glob, &si); - if (unlikely(ret != 0)) - goto out_no_zone; -#endif - for (i = 0; i < glob->num_zones; ++i) { - zone = glob->zones[i]; - pr_info("Zone %7s: Available graphics memory: %llu KiB\n", - zone->name, (unsigned long long)zone->max_mem >> 10); - } - ttm_pool_mgr_init(glob->zone_kernel->max_mem/(2*PAGE_SIZE)); - ttm_tt_mgr_init(); - return 0; -out_no_zone: - ttm_mem_global_release(glob); - return ret; -} - -void ttm_mem_global_release(struct ttm_mem_global *glob) -{ - struct ttm_mem_zone *zone; - unsigned int i; - - /* let the page allocator first stop the shrink work. */ - ttm_pool_mgr_fini(); - ttm_tt_mgr_fini(); - - flush_workqueue(glob->swap_queue); - destroy_workqueue(glob->swap_queue); - glob->swap_queue = NULL; - for (i = 0; i < glob->num_zones; ++i) { - zone = glob->zones[i]; - kobject_del(&zone->kobj); - kobject_put(&zone->kobj); - } - kobject_del(&glob->kobj); - kobject_put(&glob->kobj); - memset(glob, 0, sizeof(*glob)); -} - -static void ttm_check_swapping(struct ttm_mem_global *glob) -{ - bool needs_swapping = false; - unsigned int i; - struct ttm_mem_zone *zone; - - spin_lock(&glob->lock); - for (i = 0; i < glob->num_zones; ++i) { - zone = glob->zones[i]; - if (zone->used_mem > zone->swap_limit) { - needs_swapping = true; - break; - } - } - - spin_unlock(&glob->lock); - - if (unlikely(needs_swapping)) - (void)queue_work(glob->swap_queue, &glob->work); - -} - -static void ttm_mem_global_free_zone(struct ttm_mem_global *glob, - struct ttm_mem_zone *single_zone, - uint64_t amount) -{ - unsigned int i; - struct ttm_mem_zone *zone; - - spin_lock(&glob->lock); - for (i = 0; i < glob->num_zones; ++i) { - zone = glob->zones[i]; - if (single_zone && zone != single_zone) - continue; - zone->used_mem -= amount; - } - spin_unlock(&glob->lock); -} - -void ttm_mem_global_free(struct ttm_mem_global *glob, - uint64_t amount) -{ - return ttm_mem_global_free_zone(glob, glob->zone_kernel, amount); -} -EXPORT_SYMBOL(ttm_mem_global_free); - -/* - * check if the available mem is under lower memory limit - * - * a. if no swap disk at all or free swap space is under swap_mem_limit - * but available system mem is bigger than sys_mem_limit, allow TTM - * allocation; - * - * b. if the available system mem is less than sys_mem_limit but free - * swap disk is bigger than swap_mem_limit, allow TTM allocation. - */ -bool -ttm_check_under_lowerlimit(struct ttm_mem_global *glob, - uint64_t num_pages, - struct ttm_operation_ctx *ctx) -{ - int64_t available; - - /* We allow over commit during suspend */ - if (ctx->force_alloc) - return false; - - available = get_nr_swap_pages() + si_mem_available(); - available -= num_pages; - if (available < glob->lower_mem_limit) - return true; - - return false; -} - -static int ttm_mem_global_reserve(struct ttm_mem_global *glob, - struct ttm_mem_zone *single_zone, - uint64_t amount, bool reserve) -{ - uint64_t limit; - int ret = -ENOMEM; - unsigned int i; - struct ttm_mem_zone *zone; - - spin_lock(&glob->lock); - for (i = 0; i < glob->num_zones; ++i) { - zone = glob->zones[i]; - if (single_zone && zone != single_zone) - continue; - - limit = (capable(CAP_SYS_ADMIN)) ? - zone->emer_mem : zone->max_mem; - - if (zone->used_mem > limit) - goto out_unlock; - } - - if (reserve) { - for (i = 0; i < glob->num_zones; ++i) { - zone = glob->zones[i]; - if (single_zone && zone != single_zone) - continue; - zone->used_mem += amount; - } - } - - ret = 0; -out_unlock: - spin_unlock(&glob->lock); - ttm_check_swapping(glob); - - return ret; -} - - -static int ttm_mem_global_alloc_zone(struct ttm_mem_global *glob, - struct ttm_mem_zone *single_zone, - uint64_t memory, - struct ttm_operation_ctx *ctx) -{ - int count = TTM_MEMORY_ALLOC_RETRIES; - - while (unlikely(ttm_mem_global_reserve(glob, - single_zone, - memory, true) - != 0)) { - if (ctx->no_wait_gpu) - return -ENOMEM; - if (unlikely(count-- == 0)) - return -ENOMEM; - ttm_shrink(glob, false, memory + (memory >> 2) + 16, ctx); - } - - return 0; -} - -int ttm_mem_global_alloc(struct ttm_mem_global *glob, uint64_t memory, - struct ttm_operation_ctx *ctx) -{ - /** - * Normal allocations of kernel memory are registered in - * the kernel zone. - */ - - return ttm_mem_global_alloc_zone(glob, glob->zone_kernel, memory, ctx); -} -EXPORT_SYMBOL(ttm_mem_global_alloc); - -int ttm_mem_global_alloc_page(struct ttm_mem_global *glob, - struct page *page, uint64_t size, - struct ttm_operation_ctx *ctx) -{ - struct ttm_mem_zone *zone = NULL; - - /** - * Page allocations may be registed in a single zone - * only if highmem or !dma32. - */ - -#ifdef CONFIG_HIGHMEM - if (PageHighMem(page) && glob->zone_highmem != NULL) - zone = glob->zone_highmem; -#else - if (glob->zone_dma32 && page_to_pfn(page) > 0x00100000UL) - zone = glob->zone_kernel; -#endif - return ttm_mem_global_alloc_zone(glob, zone, size, ctx); -} - -void ttm_mem_global_free_page(struct ttm_mem_global *glob, struct page *page, - uint64_t size) -{ - struct ttm_mem_zone *zone = NULL; - -#ifdef CONFIG_HIGHMEM - if (PageHighMem(page) && glob->zone_highmem != NULL) - zone = glob->zone_highmem; -#else - if (glob->zone_dma32 && page_to_pfn(page) > 0x00100000UL) - zone = glob->zone_kernel; -#endif - ttm_mem_global_free_zone(glob, zone, size); -} - -size_t ttm_round_pot(size_t size) -{ - if ((size & (size - 1)) == 0) - return size; - else if (size > PAGE_SIZE) - return PAGE_ALIGN(size); - else { - size_t tmp_size = 4; - - while (tmp_size < size) - tmp_size <<= 1; - - return tmp_size; - } - return 0; -} -EXPORT_SYMBOL(ttm_round_pot); diff --git a/drivers/gpu/drm/ttm/ttm_pool.c b/drivers/gpu/drm/ttm/ttm_pool.c index e0617717113f..6b0f957d63d5 100644 --- a/drivers/gpu/drm/ttm/ttm_pool.c +++ b/drivers/gpu/drm/ttm/ttm_pool.c @@ -404,16 +404,10 @@ int ttm_pool_alloc(struct ttm_pool *pool, struct ttm_tt *tt, caching = pages + (1 << order); } - r = ttm_mem_global_alloc_page(&ttm_mem_glob, p, - (1 << order) * PAGE_SIZE, - ctx); - if (r) - goto error_free_page; - if (dma_addr) { r = ttm_pool_map(pool, order, p, &dma_addr); if (r) - goto error_global_free; + goto error_free_page; } num_pages -= 1 << order; @@ -427,9 +421,6 @@ int ttm_pool_alloc(struct ttm_pool *pool, struct ttm_tt *tt, return 0; -error_global_free: - ttm_mem_global_free_page(&ttm_mem_glob, p, (1 << order) * PAGE_SIZE); - error_free_page: ttm_pool_free_page(pool, tt->caching, order, p); @@ -464,8 +455,6 @@ void ttm_pool_free(struct ttm_pool *pool, struct ttm_tt *tt) order = ttm_pool_page_order(pool, p); num_pages = 1ULL << order; - ttm_mem_global_free_page(&ttm_mem_glob, p, - num_pages * PAGE_SIZE); if (tt->dma_address) ttm_pool_unmap(pool, tt->dma_address[i], num_pages); diff --git a/drivers/gpu/drm/vmwgfx/Makefile b/drivers/gpu/drm/vmwgfx/Makefile index cc4cdca7176e..8c02fa5852e7 100644 --- a/drivers/gpu/drm/vmwgfx/Makefile +++ b/drivers/gpu/drm/vmwgfx/Makefile @@ -9,7 +9,7 @@ vmwgfx-y := vmwgfx_execbuf.o vmwgfx_gmr.o vmwgfx_kms.o vmwgfx_drv.o \ vmwgfx_cotable.o vmwgfx_so.o vmwgfx_binding.o vmwgfx_msg.o \ vmwgfx_simple_resource.o vmwgfx_va.o vmwgfx_blit.o \ vmwgfx_validation.o vmwgfx_page_dirty.o vmwgfx_streamoutput.o \ - ttm_object.o ttm_lock.o + ttm_object.o ttm_lock.o ttm_memory.o vmwgfx-$(CONFIG_TRANSPARENT_HUGEPAGE) += vmwgfx_thp.o obj-$(CONFIG_DRM_VMWGFX) := vmwgfx.o diff --git a/drivers/gpu/drm/vmwgfx/ttm_memory.c b/drivers/gpu/drm/vmwgfx/ttm_memory.c new file mode 100644 index 000000000000..e972af07d029 --- /dev/null +++ b/drivers/gpu/drm/vmwgfx/ttm_memory.c @@ -0,0 +1,681 @@ +/* SPDX-License-Identifier: GPL-2.0 OR MIT */ +/************************************************************************** + * + * Copyright (c) 2006-2009 VMware, Inc., Palo Alto, CA., USA + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#define pr_fmt(fmt) "[TTM] " fmt + +#include +#include +#include +#include +#include +#include +#include + +#include +#include + +#include "ttm_memory.h" + +#define TTM_MEMORY_ALLOC_RETRIES 4 + +struct ttm_mem_global ttm_mem_glob; +EXPORT_SYMBOL(ttm_mem_glob); + +struct ttm_mem_zone { + struct kobject kobj; + struct ttm_mem_global *glob; + const char *name; + uint64_t zone_mem; + uint64_t emer_mem; + uint64_t max_mem; + uint64_t swap_limit; + uint64_t used_mem; +}; + +static struct attribute ttm_mem_sys = { + .name = "zone_memory", + .mode = S_IRUGO +}; +static struct attribute ttm_mem_emer = { + .name = "emergency_memory", + .mode = S_IRUGO | S_IWUSR +}; +static struct attribute ttm_mem_max = { + .name = "available_memory", + .mode = S_IRUGO | S_IWUSR +}; +static struct attribute ttm_mem_swap = { + .name = "swap_limit", + .mode = S_IRUGO | S_IWUSR +}; +static struct attribute ttm_mem_used = { + .name = "used_memory", + .mode = S_IRUGO +}; + +static void ttm_mem_zone_kobj_release(struct kobject *kobj) +{ + struct ttm_mem_zone *zone = + container_of(kobj, struct ttm_mem_zone, kobj); + + pr_info("Zone %7s: Used memory at exit: %llu KiB\n", + zone->name, (unsigned long long)zone->used_mem >> 10); + kfree(zone); +} + +static ssize_t ttm_mem_zone_show(struct kobject *kobj, + struct attribute *attr, + char *buffer) +{ + struct ttm_mem_zone *zone = + container_of(kobj, struct ttm_mem_zone, kobj); + uint64_t val = 0; + + spin_lock(&zone->glob->lock); + if (attr == &ttm_mem_sys) + val = zone->zone_mem; + else if (attr == &ttm_mem_emer) + val = zone->emer_mem; + else if (attr == &ttm_mem_max) + val = zone->max_mem; + else if (attr == &ttm_mem_swap) + val = zone->swap_limit; + else if (attr == &ttm_mem_used) + val = zone->used_mem; + spin_unlock(&zone->glob->lock); + + return snprintf(buffer, PAGE_SIZE, "%llu\n", + (unsigned long long) val >> 10); +} + +static void ttm_check_swapping(struct ttm_mem_global *glob); + +static ssize_t ttm_mem_zone_store(struct kobject *kobj, + struct attribute *attr, + const char *buffer, + size_t size) +{ + struct ttm_mem_zone *zone = + container_of(kobj, struct ttm_mem_zone, kobj); + int chars; + unsigned long val; + uint64_t val64; + + chars = sscanf(buffer, "%lu", &val); + if (chars == 0) + return size; + + val64 = val; + val64 <<= 10; + + spin_lock(&zone->glob->lock); + if (val64 > zone->zone_mem) + val64 = zone->zone_mem; + if (attr == &ttm_mem_emer) { + zone->emer_mem = val64; + if (zone->max_mem > val64) + zone->max_mem = val64; + } else if (attr == &ttm_mem_max) { + zone->max_mem = val64; + if (zone->emer_mem < val64) + zone->emer_mem = val64; + } else if (attr == &ttm_mem_swap) + zone->swap_limit = val64; + spin_unlock(&zone->glob->lock); + + ttm_check_swapping(zone->glob); + + return size; +} + +static struct attribute *ttm_mem_zone_attrs[] = { + &ttm_mem_sys, + &ttm_mem_emer, + &ttm_mem_max, + &ttm_mem_swap, + &ttm_mem_used, + NULL +}; + +static const struct sysfs_ops ttm_mem_zone_ops = { + .show = &ttm_mem_zone_show, + .store = &ttm_mem_zone_store +}; + +static struct kobj_type ttm_mem_zone_kobj_type = { + .release = &ttm_mem_zone_kobj_release, + .sysfs_ops = &ttm_mem_zone_ops, + .default_attrs = ttm_mem_zone_attrs, +}; + +static struct attribute ttm_mem_global_lower_mem_limit = { + .name = "lower_mem_limit", + .mode = S_IRUGO | S_IWUSR +}; + +static ssize_t ttm_mem_global_show(struct kobject *kobj, + struct attribute *attr, + char *buffer) +{ + struct ttm_mem_global *glob = + container_of(kobj, struct ttm_mem_global, kobj); + uint64_t val = 0; + + spin_lock(&glob->lock); + val = glob->lower_mem_limit; + spin_unlock(&glob->lock); + /* convert from number of pages to KB */ + val <<= (PAGE_SHIFT - 10); + return snprintf(buffer, PAGE_SIZE, "%llu\n", + (unsigned long long) val); +} + +static ssize_t ttm_mem_global_store(struct kobject *kobj, + struct attribute *attr, + const char *buffer, + size_t size) +{ + int chars; + uint64_t val64; + unsigned long val; + struct ttm_mem_global *glob = + container_of(kobj, struct ttm_mem_global, kobj); + + chars = sscanf(buffer, "%lu", &val); + if (chars == 0) + return size; + + val64 = val; + /* convert from KB to number of pages */ + val64 >>= (PAGE_SHIFT - 10); + + spin_lock(&glob->lock); + glob->lower_mem_limit = val64; + spin_unlock(&glob->lock); + + return size; +} + +static struct attribute *ttm_mem_global_attrs[] = { + &ttm_mem_global_lower_mem_limit, + NULL +}; + +static const struct sysfs_ops ttm_mem_global_ops = { + .show = &ttm_mem_global_show, + .store = &ttm_mem_global_store, +}; + +static struct kobj_type ttm_mem_glob_kobj_type = { + .sysfs_ops = &ttm_mem_global_ops, + .default_attrs = ttm_mem_global_attrs, +}; + +static bool ttm_zones_above_swap_target(struct ttm_mem_global *glob, + bool from_wq, uint64_t extra) +{ + unsigned int i; + struct ttm_mem_zone *zone; + uint64_t target; + + for (i = 0; i < glob->num_zones; ++i) { + zone = glob->zones[i]; + + if (from_wq) + target = zone->swap_limit; + else if (capable(CAP_SYS_ADMIN)) + target = zone->emer_mem; + else + target = zone->max_mem; + + target = (extra > target) ? 0ULL : target; + + if (zone->used_mem > target) + return true; + } + return false; +} + +/* + * At this point we only support a single shrink callback. + * Extend this if needed, perhaps using a linked list of callbacks. + * Note that this function is reentrant: + * many threads may try to swap out at any given time. + */ + +static void ttm_shrink(struct ttm_mem_global *glob, bool from_wq, + uint64_t extra, struct ttm_operation_ctx *ctx) +{ + int ret; + + spin_lock(&glob->lock); + + while (ttm_zones_above_swap_target(glob, from_wq, extra)) { + spin_unlock(&glob->lock); + ret = ttm_bo_swapout(ctx, GFP_KERNEL); + spin_lock(&glob->lock); + if (unlikely(ret < 0)) + break; + } + + spin_unlock(&glob->lock); +} + +static void ttm_shrink_work(struct work_struct *work) +{ + struct ttm_operation_ctx ctx = { + .interruptible = false, + .no_wait_gpu = false + }; + struct ttm_mem_global *glob = + container_of(work, struct ttm_mem_global, work); + + ttm_shrink(glob, true, 0ULL, &ctx); +} + +static int ttm_mem_init_kernel_zone(struct ttm_mem_global *glob, + const struct sysinfo *si) +{ + struct ttm_mem_zone *zone = kzalloc(sizeof(*zone), GFP_KERNEL); + uint64_t mem; + int ret; + + if (unlikely(!zone)) + return -ENOMEM; + + mem = si->totalram - si->totalhigh; + mem *= si->mem_unit; + + zone->name = "kernel"; + zone->zone_mem = mem; + zone->max_mem = mem >> 1; + zone->emer_mem = (mem >> 1) + (mem >> 2); + zone->swap_limit = zone->max_mem - (mem >> 3); + zone->used_mem = 0; + zone->glob = glob; + glob->zone_kernel = zone; + ret = kobject_init_and_add( + &zone->kobj, &ttm_mem_zone_kobj_type, &glob->kobj, zone->name); + if (unlikely(ret != 0)) { + kobject_put(&zone->kobj); + return ret; + } + glob->zones[glob->num_zones++] = zone; + return 0; +} + +#ifdef CONFIG_HIGHMEM +static int ttm_mem_init_highmem_zone(struct ttm_mem_global *glob, + const struct sysinfo *si) +{ + struct ttm_mem_zone *zone; + uint64_t mem; + int ret; + + if (si->totalhigh == 0) + return 0; + + zone = kzalloc(sizeof(*zone), GFP_KERNEL); + if (unlikely(!zone)) + return -ENOMEM; + + mem = si->totalram; + mem *= si->mem_unit; + + zone->name = "highmem"; + zone->zone_mem = mem; + zone->max_mem = mem >> 1; + zone->emer_mem = (mem >> 1) + (mem >> 2); + zone->swap_limit = zone->max_mem - (mem >> 3); + zone->used_mem = 0; + zone->glob = glob; + glob->zone_highmem = zone; + ret = kobject_init_and_add( + &zone->kobj, &ttm_mem_zone_kobj_type, &glob->kobj, "%s", + zone->name); + if (unlikely(ret != 0)) { + kobject_put(&zone->kobj); + return ret; + } + glob->zones[glob->num_zones++] = zone; + return 0; +} +#else +static int ttm_mem_init_dma32_zone(struct ttm_mem_global *glob, + const struct sysinfo *si) +{ + struct ttm_mem_zone *zone = kzalloc(sizeof(*zone), GFP_KERNEL); + uint64_t mem; + int ret; + + if (unlikely(!zone)) + return -ENOMEM; + + mem = si->totalram; + mem *= si->mem_unit; + + /** + * No special dma32 zone needed. + */ + + if (mem <= ((uint64_t) 1ULL << 32)) { + kfree(zone); + return 0; + } + + /* + * Limit max dma32 memory to 4GB for now + * until we can figure out how big this + * zone really is. + */ + + mem = ((uint64_t) 1ULL << 32); + zone->name = "dma32"; + zone->zone_mem = mem; + zone->max_mem = mem >> 1; + zone->emer_mem = (mem >> 1) + (mem >> 2); + zone->swap_limit = zone->max_mem - (mem >> 3); + zone->used_mem = 0; + zone->glob = glob; + glob->zone_dma32 = zone; + ret = kobject_init_and_add( + &zone->kobj, &ttm_mem_zone_kobj_type, &glob->kobj, zone->name); + if (unlikely(ret != 0)) { + kobject_put(&zone->kobj); + return ret; + } + glob->zones[glob->num_zones++] = zone; + return 0; +} +#endif + +int ttm_mem_global_init(struct ttm_mem_global *glob, struct device *dev) +{ + struct sysinfo si; + int ret; + int i; + struct ttm_mem_zone *zone; + + spin_lock_init(&glob->lock); + glob->swap_queue = create_singlethread_workqueue("ttm_swap"); + INIT_WORK(&glob->work, ttm_shrink_work); + + ret = kobject_init_and_add(&glob->kobj, &ttm_mem_glob_kobj_type, + &dev->kobj, "memory_accounting"); + if (unlikely(ret != 0)) { + kobject_put(&glob->kobj); + return ret; + } + + si_meminfo(&si); + + /* set it as 0 by default to keep original behavior of OOM */ + glob->lower_mem_limit = 0; + + ret = ttm_mem_init_kernel_zone(glob, &si); + if (unlikely(ret != 0)) + goto out_no_zone; +#ifdef CONFIG_HIGHMEM + ret = ttm_mem_init_highmem_zone(glob, &si); + if (unlikely(ret != 0)) + goto out_no_zone; +#else + ret = ttm_mem_init_dma32_zone(glob, &si); + if (unlikely(ret != 0)) + goto out_no_zone; +#endif + for (i = 0; i < glob->num_zones; ++i) { + zone = glob->zones[i]; + pr_info("Zone %7s: Available graphics memory: %llu KiB\n", + zone->name, (unsigned long long)zone->max_mem >> 10); + } + return 0; +out_no_zone: + ttm_mem_global_release(glob); + return ret; +} + +void ttm_mem_global_release(struct ttm_mem_global *glob) +{ + struct ttm_mem_zone *zone; + unsigned int i; + + flush_workqueue(glob->swap_queue); + destroy_workqueue(glob->swap_queue); + glob->swap_queue = NULL; + for (i = 0; i < glob->num_zones; ++i) { + zone = glob->zones[i]; + kobject_del(&zone->kobj); + kobject_put(&zone->kobj); + } + kobject_del(&glob->kobj); + kobject_put(&glob->kobj); + memset(glob, 0, sizeof(*glob)); +} + +static void ttm_check_swapping(struct ttm_mem_global *glob) +{ + bool needs_swapping = false; + unsigned int i; + struct ttm_mem_zone *zone; + + spin_lock(&glob->lock); + for (i = 0; i < glob->num_zones; ++i) { + zone = glob->zones[i]; + if (zone->used_mem > zone->swap_limit) { + needs_swapping = true; + break; + } + } + + spin_unlock(&glob->lock); + + if (unlikely(needs_swapping)) + (void)queue_work(glob->swap_queue, &glob->work); + +} + +static void ttm_mem_global_free_zone(struct ttm_mem_global *glob, + struct ttm_mem_zone *single_zone, + uint64_t amount) +{ + unsigned int i; + struct ttm_mem_zone *zone; + + spin_lock(&glob->lock); + for (i = 0; i < glob->num_zones; ++i) { + zone = glob->zones[i]; + if (single_zone && zone != single_zone) + continue; + zone->used_mem -= amount; + } + spin_unlock(&glob->lock); +} + +void ttm_mem_global_free(struct ttm_mem_global *glob, + uint64_t amount) +{ + return ttm_mem_global_free_zone(glob, glob->zone_kernel, amount); +} +EXPORT_SYMBOL(ttm_mem_global_free); + +/* + * check if the available mem is under lower memory limit + * + * a. if no swap disk at all or free swap space is under swap_mem_limit + * but available system mem is bigger than sys_mem_limit, allow TTM + * allocation; + * + * b. if the available system mem is less than sys_mem_limit but free + * swap disk is bigger than swap_mem_limit, allow TTM allocation. + */ +bool +ttm_check_under_lowerlimit(struct ttm_mem_global *glob, + uint64_t num_pages, + struct ttm_operation_ctx *ctx) +{ + int64_t available; + + /* We allow over commit during suspend */ + if (ctx->force_alloc) + return false; + + available = get_nr_swap_pages() + si_mem_available(); + available -= num_pages; + if (available < glob->lower_mem_limit) + return true; + + return false; +} + +static int ttm_mem_global_reserve(struct ttm_mem_global *glob, + struct ttm_mem_zone *single_zone, + uint64_t amount, bool reserve) +{ + uint64_t limit; + int ret = -ENOMEM; + unsigned int i; + struct ttm_mem_zone *zone; + + spin_lock(&glob->lock); + for (i = 0; i < glob->num_zones; ++i) { + zone = glob->zones[i]; + if (single_zone && zone != single_zone) + continue; + + limit = (capable(CAP_SYS_ADMIN)) ? + zone->emer_mem : zone->max_mem; + + if (zone->used_mem > limit) + goto out_unlock; + } + + if (reserve) { + for (i = 0; i < glob->num_zones; ++i) { + zone = glob->zones[i]; + if (single_zone && zone != single_zone) + continue; + zone->used_mem += amount; + } + } + + ret = 0; +out_unlock: + spin_unlock(&glob->lock); + ttm_check_swapping(glob); + + return ret; +} + + +static int ttm_mem_global_alloc_zone(struct ttm_mem_global *glob, + struct ttm_mem_zone *single_zone, + uint64_t memory, + struct ttm_operation_ctx *ctx) +{ + int count = TTM_MEMORY_ALLOC_RETRIES; + + while (unlikely(ttm_mem_global_reserve(glob, + single_zone, + memory, true) + != 0)) { + if (ctx->no_wait_gpu) + return -ENOMEM; + if (unlikely(count-- == 0)) + return -ENOMEM; + ttm_shrink(glob, false, memory + (memory >> 2) + 16, ctx); + } + + return 0; +} + +int ttm_mem_global_alloc(struct ttm_mem_global *glob, uint64_t memory, + struct ttm_operation_ctx *ctx) +{ + /** + * Normal allocations of kernel memory are registered in + * the kernel zone. + */ + + return ttm_mem_global_alloc_zone(glob, glob->zone_kernel, memory, ctx); +} +EXPORT_SYMBOL(ttm_mem_global_alloc); + +int ttm_mem_global_alloc_page(struct ttm_mem_global *glob, + struct page *page, uint64_t size, + struct ttm_operation_ctx *ctx) +{ + struct ttm_mem_zone *zone = NULL; + + /** + * Page allocations may be registed in a single zone + * only if highmem or !dma32. + */ + +#ifdef CONFIG_HIGHMEM + if (PageHighMem(page) && glob->zone_highmem != NULL) + zone = glob->zone_highmem; +#else + if (glob->zone_dma32 && page_to_pfn(page) > 0x00100000UL) + zone = glob->zone_kernel; +#endif + return ttm_mem_global_alloc_zone(glob, zone, size, ctx); +} + +void ttm_mem_global_free_page(struct ttm_mem_global *glob, struct page *page, + uint64_t size) +{ + struct ttm_mem_zone *zone = NULL; + +#ifdef CONFIG_HIGHMEM + if (PageHighMem(page) && glob->zone_highmem != NULL) + zone = glob->zone_highmem; +#else + if (glob->zone_dma32 && page_to_pfn(page) > 0x00100000UL) + zone = glob->zone_kernel; +#endif + ttm_mem_global_free_zone(glob, zone, size); +} + +size_t ttm_round_pot(size_t size) +{ + if ((size & (size - 1)) == 0) + return size; + else if (size > PAGE_SIZE) + return PAGE_ALIGN(size); + else { + size_t tmp_size = 4; + + while (tmp_size < size) + tmp_size <<= 1; + + return tmp_size; + } + return 0; +} +EXPORT_SYMBOL(ttm_round_pot); diff --git a/drivers/gpu/drm/vmwgfx/ttm_memory.h b/drivers/gpu/drm/vmwgfx/ttm_memory.h new file mode 100644 index 000000000000..c50dba774485 --- /dev/null +++ b/drivers/gpu/drm/vmwgfx/ttm_memory.h @@ -0,0 +1,96 @@ +/************************************************************************** + * + * Copyright (c) 2006-2009 VMware, Inc., Palo Alto, CA., USA + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#ifndef TTM_MEMORY_H +#define TTM_MEMORY_H + +#include +#include +#include +#include +#include +#include +#include + +#include + +/** + * struct ttm_mem_global - Global memory accounting structure. + * + * @shrink: A single callback to shrink TTM memory usage. Extend this + * to a linked list to be able to handle multiple callbacks when needed. + * @swap_queue: A workqueue to handle shrinking in low memory situations. We + * need a separate workqueue since it will spend a lot of time waiting + * for the GPU, and this will otherwise block other workqueue tasks(?) + * At this point we use only a single-threaded workqueue. + * @work: The workqueue callback for the shrink queue. + * @lock: Lock to protect the @shrink - and the memory accounting members, + * that is, essentially the whole structure with some exceptions. + * @lower_mem_limit: include lower limit of swap space and lower limit of + * system memory. + * @zones: Array of pointers to accounting zones. + * @num_zones: Number of populated entries in the @zones array. + * @zone_kernel: Pointer to the kernel zone. + * @zone_highmem: Pointer to the highmem zone if there is one. + * @zone_dma32: Pointer to the dma32 zone if there is one. + * + * Note that this structure is not per device. It should be global for all + * graphics devices. + */ + +#define TTM_MEM_MAX_ZONES 2 +struct ttm_mem_zone; +extern struct ttm_mem_global { + struct kobject kobj; + struct workqueue_struct *swap_queue; + struct work_struct work; + spinlock_t lock; + uint64_t lower_mem_limit; + struct ttm_mem_zone *zones[TTM_MEM_MAX_ZONES]; + unsigned int num_zones; + struct ttm_mem_zone *zone_kernel; +#ifdef CONFIG_HIGHMEM + struct ttm_mem_zone *zone_highmem; +#else + struct ttm_mem_zone *zone_dma32; +#endif +} ttm_mem_glob; + +int ttm_mem_global_init(struct ttm_mem_global *glob, struct device *dev); +void ttm_mem_global_release(struct ttm_mem_global *glob); +int ttm_mem_global_alloc(struct ttm_mem_global *glob, uint64_t memory, + struct ttm_operation_ctx *ctx); +void ttm_mem_global_free(struct ttm_mem_global *glob, uint64_t amount); +int ttm_mem_global_alloc_page(struct ttm_mem_global *glob, + struct page *page, uint64_t size, + struct ttm_operation_ctx *ctx); +void ttm_mem_global_free_page(struct ttm_mem_global *glob, + struct page *page, uint64_t size); +size_t ttm_round_pot(size_t size); +bool ttm_check_under_lowerlimit(struct ttm_mem_global *glob, uint64_t num_pages, + struct ttm_operation_ctx *ctx); +#endif diff --git a/drivers/gpu/drm/vmwgfx/ttm_object.h b/drivers/gpu/drm/vmwgfx/ttm_object.h index ede26df87c93..49b064f0cb19 100644 --- a/drivers/gpu/drm/vmwgfx/ttm_object.h +++ b/drivers/gpu/drm/vmwgfx/ttm_object.h @@ -43,7 +43,8 @@ #include #include -#include + +#include "ttm_memory.h" /** * enum ttm_ref_type diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_bo.c b/drivers/gpu/drm/vmwgfx/vmwgfx_bo.c index 6b3bfd8c678a..50e529a01677 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_bo.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_bo.c @@ -507,11 +507,16 @@ int vmw_bo_create_kernel(struct vmw_private *dev_priv, unsigned long size, acc_size = ttm_round_pot(sizeof(*bo)); acc_size += ttm_round_pot(npages * sizeof(void *)); acc_size += ttm_round_pot(sizeof(struct ttm_tt)); + + ret = ttm_mem_global_alloc(&ttm_mem_glob, acc_size, &ctx); + if (unlikely(ret)) + goto error_free; + ret = ttm_bo_init_reserved(&dev_priv->bdev, bo, size, ttm_bo_type_device, placement, 0, - &ctx, acc_size, NULL, NULL, NULL); + &ctx, NULL, NULL, NULL); if (unlikely(ret)) - goto error_free; + goto error_account; ttm_bo_pin(bo); ttm_bo_unreserve(bo); @@ -519,6 +524,9 @@ int vmw_bo_create_kernel(struct vmw_private *dev_priv, unsigned long size, return 0; +error_account: + ttm_mem_global_free(&ttm_mem_glob, acc_size); + error_free: kfree(bo); return ret; @@ -558,11 +566,17 @@ int vmw_bo_init(struct vmw_private *dev_priv, vmw_bo->base.priority = 3; vmw_bo->res_tree = RB_ROOT; + ret = ttm_mem_global_alloc(&ttm_mem_glob, acc_size, &ctx); + if (unlikely(ret)) + return ret; + ret = ttm_bo_init_reserved(bdev, &vmw_bo->base, size, ttm_bo_type_device, placement, - 0, &ctx, acc_size, NULL, NULL, bo_free); - if (unlikely(ret)) + 0, &ctx, NULL, NULL, bo_free); + if (unlikely(ret)) { + ttm_mem_global_free(&ttm_mem_glob, acc_size); return ret; + } if (pin) ttm_bo_pin(&vmw_bo->base); diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c index 9a9afa087802..4efed3bf0ef9 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c @@ -1268,6 +1268,7 @@ static void vmw_remove(struct pci_dev *pdev) { struct drm_device *dev = pci_get_drvdata(pdev); + ttm_mem_global_release(&ttm_mem_glob); drm_dev_unregister(dev); vmw_driver_unload(dev); } @@ -1518,6 +1519,10 @@ static int vmw_probe(struct pci_dev *pdev, const struct pci_device_id *ent) pci_set_drvdata(pdev, &vmw->drm); + ret = ttm_mem_global_init(&ttm_mem_glob, &pdev->dev); + if (ret) + return ret; + ret = vmw_driver_load(vmw, ent->device); if (ret) return ret; diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_ttm_buffer.c b/drivers/gpu/drm/vmwgfx/vmwgfx_ttm_buffer.c index d1bfa59579f1..63f10c865061 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_ttm_buffer.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_ttm_buffer.c @@ -576,11 +576,31 @@ static void vmw_ttm_destroy(struct ttm_device *bdev, struct ttm_tt *ttm) static int vmw_ttm_populate(struct ttm_device *bdev, struct ttm_tt *ttm, struct ttm_operation_ctx *ctx) { + unsigned int i; + int ret; + /* TODO: maybe completely drop this ? */ if (ttm_tt_is_populated(ttm)) return 0; - return ttm_pool_alloc(&bdev->pool, ttm, ctx); + ret = ttm_pool_alloc(&bdev->pool, ttm, ctx); + if (ret) + return ret; + + for (i = 0; i < ttm->num_pages; ++i) { + ret = ttm_mem_global_alloc_page(&ttm_mem_glob, ttm->pages[i], + PAGE_SIZE, ctx); + if (ret) + goto error; + } + return 0; + +error: + while (i--) + ttm_mem_global_free_page(&ttm_mem_glob, ttm->pages[i], + PAGE_SIZE); + ttm_pool_free(&bdev->pool, ttm); + return ret; } static void vmw_ttm_unpopulate(struct ttm_device *bdev, @@ -588,6 +608,7 @@ static void vmw_ttm_unpopulate(struct ttm_device *bdev, { struct vmw_ttm_tt *vmw_tt = container_of(ttm, struct vmw_ttm_tt, dma_ttm); + unsigned int i; if (vmw_tt->mob) { vmw_mob_destroy(vmw_tt->mob); @@ -595,6 +616,11 @@ static void vmw_ttm_unpopulate(struct ttm_device *bdev, } vmw_ttm_unmap_dma(vmw_tt); + + for (i = 0; i < ttm->num_pages; ++i) + ttm_mem_global_free_page(&ttm_mem_glob, ttm->pages[i], + PAGE_SIZE); + ttm_pool_free(&bdev->pool, ttm); } diff --git a/include/drm/ttm/ttm_bo_api.h b/include/drm/ttm/ttm_bo_api.h index 1297a8fb7ccb..4fb523dfab32 100644 --- a/include/drm/ttm/ttm_bo_api.h +++ b/include/drm/ttm/ttm_bo_api.h @@ -88,7 +88,6 @@ struct ttm_tt; * @type: The bo type. * @destroy: Destruction function. If NULL, kfree is used. * @num_pages: Actual number of pages. - * @acc_size: Accounted size for this object. * @kref: Reference count of this buffer object. When this refcount reaches * zero, the object is destroyed or put on the delayed delete list. * @mem: structure describing current placement. @@ -125,7 +124,6 @@ struct ttm_buffer_object { struct ttm_device *bdev; enum ttm_bo_type type; void (*destroy) (struct ttm_buffer_object *); - size_t acc_size; /** * Members not needing protection. @@ -357,10 +355,6 @@ void ttm_bo_unlock_delayed_workqueue(struct ttm_device *bdev, int resched); bool ttm_bo_eviction_valuable(struct ttm_buffer_object *bo, const struct ttm_place *place); -size_t ttm_bo_dma_acc_size(struct ttm_device *bdev, - unsigned long bo_size, - unsigned struct_size); - /** * ttm_bo_init_reserved * @@ -371,7 +365,6 @@ size_t ttm_bo_dma_acc_size(struct ttm_device *bdev, * @flags: Initial placement flags. * @page_alignment: Data alignment in pages. * @ctx: TTM operation context for memory allocation. - * @acc_size: Accounted size for this object. * @resv: Pointer to a dma_resv, or NULL to let ttm allocate one. * @destroy: Destroy function. Use NULL for kfree(). * @@ -402,8 +395,7 @@ int ttm_bo_init_reserved(struct ttm_device *bdev, struct ttm_placement *placement, uint32_t page_alignment, struct ttm_operation_ctx *ctx, - size_t acc_size, struct sg_table *sg, - struct dma_resv *resv, + struct sg_table *sg, struct dma_resv *resv, void (*destroy) (struct ttm_buffer_object *)); /** @@ -421,7 +413,6 @@ int ttm_bo_init_reserved(struct ttm_device *bdev, * holds a pointer to a persistent shmem object. Typically, this would * point to the shmem object backing a GEM object if TTM is used to back a * GEM user interface. - * @acc_size: Accounted size for this object. * @resv: Pointer to a dma_resv, or NULL to let ttm allocate one. * @destroy: Destroy function. Use NULL for kfree(). * @@ -446,7 +437,7 @@ int ttm_bo_init_reserved(struct ttm_device *bdev, int ttm_bo_init(struct ttm_device *bdev, struct ttm_buffer_object *bo, size_t size, enum ttm_bo_type type, struct ttm_placement *placement, - uint32_t page_alignment, bool interrubtible, size_t acc_size, + uint32_t page_alignment, bool interrubtible, struct sg_table *sg, struct dma_resv *resv, void (*destroy) (struct ttm_buffer_object *)); diff --git a/include/drm/ttm/ttm_bo_driver.h b/include/drm/ttm/ttm_bo_driver.h index 1c9bf993e252..8959c0075cfd 100644 --- a/include/drm/ttm/ttm_bo_driver.h +++ b/include/drm/ttm/ttm_bo_driver.h @@ -40,7 +40,6 @@ #include #include "ttm_bo_api.h" -#include "ttm_memory.h" #include "ttm_placement.h" #include "ttm_tt.h" #include "ttm_pool.h" diff --git a/include/drm/ttm/ttm_memory.h b/include/drm/ttm/ttm_memory.h deleted file mode 100644 index c1f167881e33..000000000000 --- a/include/drm/ttm/ttm_memory.h +++ /dev/null @@ -1,95 +0,0 @@ -/************************************************************************** - * - * Copyright (c) 2006-2009 VMware, Inc., Palo Alto, CA., USA - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, - * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR - * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE - * USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - -#ifndef TTM_MEMORY_H -#define TTM_MEMORY_H - -#include -#include -#include -#include -#include -#include -#include -#include "ttm_bo_api.h" - -/** - * struct ttm_mem_global - Global memory accounting structure. - * - * @shrink: A single callback to shrink TTM memory usage. Extend this - * to a linked list to be able to handle multiple callbacks when needed. - * @swap_queue: A workqueue to handle shrinking in low memory situations. We - * need a separate workqueue since it will spend a lot of time waiting - * for the GPU, and this will otherwise block other workqueue tasks(?) - * At this point we use only a single-threaded workqueue. - * @work: The workqueue callback for the shrink queue. - * @lock: Lock to protect the @shrink - and the memory accounting members, - * that is, essentially the whole structure with some exceptions. - * @lower_mem_limit: include lower limit of swap space and lower limit of - * system memory. - * @zones: Array of pointers to accounting zones. - * @num_zones: Number of populated entries in the @zones array. - * @zone_kernel: Pointer to the kernel zone. - * @zone_highmem: Pointer to the highmem zone if there is one. - * @zone_dma32: Pointer to the dma32 zone if there is one. - * - * Note that this structure is not per device. It should be global for all - * graphics devices. - */ - -#define TTM_MEM_MAX_ZONES 2 -struct ttm_mem_zone; -extern struct ttm_mem_global { - struct kobject kobj; - struct workqueue_struct *swap_queue; - struct work_struct work; - spinlock_t lock; - uint64_t lower_mem_limit; - struct ttm_mem_zone *zones[TTM_MEM_MAX_ZONES]; - unsigned int num_zones; - struct ttm_mem_zone *zone_kernel; -#ifdef CONFIG_HIGHMEM - struct ttm_mem_zone *zone_highmem; -#else - struct ttm_mem_zone *zone_dma32; -#endif -} ttm_mem_glob; - -int ttm_mem_global_init(struct ttm_mem_global *glob); -void ttm_mem_global_release(struct ttm_mem_global *glob); -int ttm_mem_global_alloc(struct ttm_mem_global *glob, uint64_t memory, - struct ttm_operation_ctx *ctx); -void ttm_mem_global_free(struct ttm_mem_global *glob, uint64_t amount); -int ttm_mem_global_alloc_page(struct ttm_mem_global *glob, - struct page *page, uint64_t size, - struct ttm_operation_ctx *ctx); -void ttm_mem_global_free_page(struct ttm_mem_global *glob, - struct page *page, uint64_t size); -size_t ttm_round_pot(size_t size); -bool ttm_check_under_lowerlimit(struct ttm_mem_global *glob, uint64_t num_pages, - struct ttm_operation_ctx *ctx); -#endif diff --git a/include/drm/ttm/ttm_tt.h b/include/drm/ttm/ttm_tt.h index cce57fb49e2c..069f8130241a 100644 --- a/include/drm/ttm/ttm_tt.h +++ b/include/drm/ttm/ttm_tt.h @@ -30,6 +30,7 @@ #include #include +struct ttm_bo_device; struct ttm_tt; struct ttm_resource; struct ttm_buffer_object; -- cgit v1.2.3 From 817eab2fdc86c4d71c6401c72f25165db6b06dcd Mon Sep 17 00:00:00 2001 From: Zack Rusin Date: Wed, 17 Feb 2021 13:55:21 -0500 Subject: drm/vmwgfx: Bump the minor version and cleanup logging We should advertise the reworked fifo/cmd handling and huge pages support. Also while bumping the version number lets cleanup the logging, there's no point in logging whether we're atomic (we always are) or what repo we use (it's always in kernel now). Signed-off-by: Zack Rusin Reviewed-by: Martin Krastev Reviewed-by: Roland Scheidegger Link: https://patchwork.freedesktop.org/patch/msgid/20210224163034.918430-1-zackr@vmware.com --- drivers/gpu/drm/vmwgfx/vmwgfx_drv.c | 13 ------------- drivers/gpu/drm/vmwgfx/vmwgfx_drv.h | 4 ++-- 2 files changed, 2 insertions(+), 15 deletions(-) (limited to 'drivers/gpu/drm/vmwgfx/vmwgfx_drv.c') diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c index 4efed3bf0ef9..6910111099c8 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c @@ -47,12 +47,6 @@ #define VMW_MIN_INITIAL_WIDTH 800 #define VMW_MIN_INITIAL_HEIGHT 600 -#ifndef VMWGFX_GIT_VERSION -#define VMWGFX_GIT_VERSION "Unknown" -#endif - -#define VMWGFX_REPO "In Tree" - #define VMWGFX_VALIDATION_MEM_GRAN (16*PAGE_SIZE) @@ -967,8 +961,6 @@ static int vmw_driver_load(struct vmw_private *dev_priv, u32 pci_id) if (ret) goto out_no_fifo; - DRM_INFO("Atomic: %s\n", (dev_priv->drm.driver->driver_features & DRIVER_ATOMIC) - ? "yes." : "no."); if (dev_priv->sm_type == VMW_SM_5) DRM_INFO("SM5 support available.\n"); if (dev_priv->sm_type == VMW_SM_4_1) @@ -976,11 +968,6 @@ static int vmw_driver_load(struct vmw_private *dev_priv, u32 pci_id) if (dev_priv->sm_type == VMW_SM_4) DRM_INFO("SM4 support available.\n"); - snprintf(host_log, sizeof(host_log), "vmwgfx: %s-%s", - VMWGFX_REPO, VMWGFX_GIT_VERSION); - vmw_host_log(host_log); - - memset(host_log, 0, sizeof(host_log)); snprintf(host_log, sizeof(host_log), "vmwgfx: Module Version: %d.%d.%d", VMWGFX_DRIVER_MAJOR, VMWGFX_DRIVER_MINOR, VMWGFX_DRIVER_PATCHLEVEL); diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h index e65b00f8336d..bb2ce6327944 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h @@ -55,10 +55,10 @@ #define VMWGFX_DRIVER_NAME "vmwgfx" -#define VMWGFX_DRIVER_DATE "20200114" +#define VMWGFX_DRIVER_DATE "20210218" #define VMWGFX_DRIVER_MAJOR 2 #define VMWGFX_DRIVER_MINOR 18 -#define VMWGFX_DRIVER_PATCHLEVEL 0 +#define VMWGFX_DRIVER_PATCHLEVEL 1 #define VMWGFX_FIFO_STATIC_SIZE (1024*1024) #define VMWGFX_MAX_RELOCATIONS 2048 #define VMWGFX_MAX_VALIDATIONS 2048 -- cgit v1.2.3 From ebd59851c796c221daf0d3b594fb2533f87161cf Mon Sep 17 00:00:00 2001 From: Christian König Date: Tue, 6 Oct 2020 13:35:32 +0200 Subject: drm/ttm: move swapout logic around v3 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Move the iteration of the global lru into the new function ttm_global_swapout() and use that instead in drivers. v2: consistently return int v3: fix build fail Signed-off-by: Christian König Tested-by: Nirmoy Das Reviewed-by: Huang Rui Reviewed-by: Matthew Auld Link: https://patchwork.freedesktop.org/patch/424008/ --- drivers/gpu/drm/ttm/ttm_bo.c | 57 +++++++++++-------------------------- drivers/gpu/drm/ttm/ttm_device.c | 29 +++++++++++++++++++ drivers/gpu/drm/ttm/ttm_tt.c | 2 +- drivers/gpu/drm/vmwgfx/ttm_memory.c | 3 +- drivers/gpu/drm/vmwgfx/vmwgfx_drv.c | 2 +- include/drm/ttm/ttm_bo_api.h | 3 +- include/drm/ttm/ttm_device.h | 2 ++ 7 files changed, 53 insertions(+), 45 deletions(-) (limited to 'drivers/gpu/drm/vmwgfx/vmwgfx_drv.c') diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index 3c23e863a3f0..66e00b404ec3 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -1193,56 +1193,35 @@ int ttm_bo_wait(struct ttm_buffer_object *bo, } EXPORT_SYMBOL(ttm_bo_wait); -/* - * A buffer object shrink method that tries to swap out the first - * buffer object on the bo_global::swap_lru list. - */ -int ttm_bo_swapout(struct ttm_operation_ctx *ctx, gfp_t gfp_flags) +int ttm_bo_swapout(struct ttm_buffer_object *bo, struct ttm_operation_ctx *ctx, + gfp_t gfp_flags) { struct ttm_global *glob = &ttm_glob; - struct ttm_buffer_object *bo; - int ret = -EBUSY; bool locked; - unsigned i; - - spin_lock(&glob->lru_lock); - for (i = 0; i < TTM_MAX_BO_PRIORITY; ++i) { - list_for_each_entry(bo, &glob->swap_lru[i], swap) { - if (!ttm_bo_evict_swapout_allowable(bo, ctx, &locked, - NULL)) - continue; - - if (!ttm_bo_get_unless_zero(bo)) { - if (locked) - dma_resv_unlock(bo->base.resv); - continue; - } + int ret; - ret = 0; - break; - } - if (!ret) - break; - } + if (!ttm_bo_evict_swapout_allowable(bo, ctx, &locked, NULL)) + return -EBUSY; - if (ret) { - spin_unlock(&glob->lru_lock); - return ret; + if (!ttm_bo_get_unless_zero(bo)) { + if (locked) + dma_resv_unlock(bo->base.resv); + return -EBUSY; } if (bo->deleted) { - ret = ttm_bo_cleanup_refs(bo, false, false, locked); + ttm_bo_cleanup_refs(bo, false, false, locked); ttm_bo_put(bo); - return ret; + return 0; } ttm_bo_del_from_lru(bo); + /* TODO: Cleanup the locking */ spin_unlock(&glob->lru_lock); - /** + /* * Move to system cached */ - if (bo->mem.mem_type != TTM_PL_SYSTEM) { struct ttm_operation_ctx ctx = { false, false }; struct ttm_resource evict_mem; @@ -1262,29 +1241,26 @@ int ttm_bo_swapout(struct ttm_operation_ctx *ctx, gfp_t gfp_flags) } } - /** + /* * Make sure BO is idle. */ - ret = ttm_bo_wait(bo, false, false); if (unlikely(ret != 0)) goto out; ttm_bo_unmap_virtual(bo); - /** + /* * Swap out. Buffer will be swapped in again as soon as * anyone tries to access a ttm page. */ - if (bo->bdev->funcs->swap_notify) bo->bdev->funcs->swap_notify(bo); ret = ttm_tt_swapout(bo->bdev, bo->ttm, gfp_flags); out: - /** - * + /* * Unreserve without putting on LRU to avoid swapping out an * already swapped buffer. */ @@ -1293,7 +1269,6 @@ out: ttm_bo_put(bo); return ret; } -EXPORT_SYMBOL(ttm_bo_swapout); void ttm_bo_tt_destroy(struct ttm_buffer_object *bo) { diff --git a/drivers/gpu/drm/ttm/ttm_device.c b/drivers/gpu/drm/ttm/ttm_device.c index 95e1b7b1f2e6..b1424189fdfb 100644 --- a/drivers/gpu/drm/ttm/ttm_device.c +++ b/drivers/gpu/drm/ttm/ttm_device.c @@ -102,6 +102,35 @@ out: return ret; } +/** + * A buffer object shrink method that tries to swap out the first + * buffer object on the global::swap_lru list. + */ +int ttm_global_swapout(struct ttm_operation_ctx *ctx, gfp_t gfp_flags) +{ + struct ttm_global *glob = &ttm_glob; + struct ttm_buffer_object *bo; + unsigned i; + int ret; + + spin_lock(&glob->lru_lock); + for (i = 0; i < TTM_MAX_BO_PRIORITY; ++i) { + list_for_each_entry(bo, &glob->swap_lru[i], swap) { + uint32_t num_pages = bo->ttm->num_pages; + + ret = ttm_bo_swapout(bo, ctx, gfp_flags); + /* ttm_bo_swapout has dropped the lru_lock */ + if (!ret) + return num_pages; + if (ret != -EBUSY) + return ret; + } + } + spin_unlock(&glob->lru_lock); + return 0; +} +EXPORT_SYMBOL(ttm_global_swapout); + static void ttm_init_sysman(struct ttm_device *bdev) { struct ttm_resource_manager *man = &bdev->sysman; diff --git a/drivers/gpu/drm/ttm/ttm_tt.c b/drivers/gpu/drm/ttm/ttm_tt.c index 2f0833c98d2c..95b5cff25f4c 100644 --- a/drivers/gpu/drm/ttm/ttm_tt.c +++ b/drivers/gpu/drm/ttm/ttm_tt.c @@ -369,7 +369,7 @@ static unsigned long ttm_tt_shrinker_scan(struct shrinker *shrink, }; int ret; - ret = ttm_bo_swapout(&ctx, GFP_NOFS); + ret = ttm_global_swapout(&ctx, GFP_NOFS); return ret < 0 ? SHRINK_EMPTY : ret; } diff --git a/drivers/gpu/drm/vmwgfx/ttm_memory.c b/drivers/gpu/drm/vmwgfx/ttm_memory.c index e972af07d029..104b95a8c7a2 100644 --- a/drivers/gpu/drm/vmwgfx/ttm_memory.c +++ b/drivers/gpu/drm/vmwgfx/ttm_memory.c @@ -38,6 +38,7 @@ #include #include +#include #include "ttm_memory.h" @@ -277,7 +278,7 @@ static void ttm_shrink(struct ttm_mem_global *glob, bool from_wq, while (ttm_zones_above_swap_target(glob, from_wq, extra)) { spin_unlock(&glob->lock); - ret = ttm_bo_swapout(ctx, GFP_KERNEL); + ret = ttm_global_swapout(ctx, GFP_KERNEL); spin_lock(&glob->lock); if (unlikely(ret < 0)) break; diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c index 6910111099c8..b991422e156c 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c @@ -1371,7 +1371,7 @@ static int vmw_pm_freeze(struct device *kdev) vmw_execbuf_release_pinned_bo(dev_priv); vmw_resource_evict_all(dev_priv); vmw_release_device_early(dev_priv); - while (ttm_bo_swapout(&ctx, GFP_KERNEL) > 0); + while (ttm_global_swapout(&ctx, GFP_KERNEL) > 0); if (dev_priv->enable_fb) vmw_fifo_resource_dec(dev_priv); if (atomic_read(&dev_priv->num_fifo_resources) != 0) { diff --git a/include/drm/ttm/ttm_bo_api.h b/include/drm/ttm/ttm_bo_api.h index 4fb523dfab32..5044ac330858 100644 --- a/include/drm/ttm/ttm_bo_api.h +++ b/include/drm/ttm/ttm_bo_api.h @@ -560,7 +560,8 @@ ssize_t ttm_bo_io(struct ttm_device *bdev, struct file *filp, const char __user *wbuf, char __user *rbuf, size_t count, loff_t *f_pos, bool write); -int ttm_bo_swapout(struct ttm_operation_ctx *ctx, gfp_t gfp_flags); +int ttm_bo_swapout(struct ttm_buffer_object *bo, struct ttm_operation_ctx *ctx, + gfp_t gfp_flags); /** * ttm_bo_uses_embedded_gem_object - check if the given bo uses the diff --git a/include/drm/ttm/ttm_device.h b/include/drm/ttm/ttm_device.h index 035bbc044a3b..ee6bba7fab6f 100644 --- a/include/drm/ttm/ttm_device.h +++ b/include/drm/ttm/ttm_device.h @@ -297,6 +297,8 @@ struct ttm_device { struct delayed_work wq; }; +int ttm_global_swapout(struct ttm_operation_ctx *ctx, gfp_t gfp_flags); + static inline struct ttm_resource_manager * ttm_manager_type(struct ttm_device *bdev, int mem_type) { -- cgit v1.2.3 From f9e2a03e110ad0c78e69201f59d18dc1c487efac Mon Sep 17 00:00:00 2001 From: Christian König Date: Tue, 6 Oct 2020 16:30:09 +0200 Subject: drm/ttm: remove swap LRU v3 MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Instead evict round robin from each devices SYSTEM and TT domain. v2: reorder num_pages access reported by Dan's script v3: fix rebase fallout, num_pages should be 32bit Signed-off-by: Christian König Tested-by: Nirmoy Das Reviewed-by: Huang Rui Reviewed-by: Matthew Auld Link: https://patchwork.freedesktop.org/patch/424009/ --- drivers/gpu/drm/ttm/ttm_bo.c | 29 ------------------ drivers/gpu/drm/ttm/ttm_bo_util.c | 1 - drivers/gpu/drm/ttm/ttm_device.c | 60 +++++++++++++++++++++++++++---------- drivers/gpu/drm/vmwgfx/vmwgfx_drv.c | 2 +- include/drm/ttm/ttm_bo_api.h | 1 - include/drm/ttm/ttm_bo_driver.h | 1 - include/drm/ttm/ttm_device.h | 7 ++--- 7 files changed, 48 insertions(+), 53 deletions(-) (limited to 'drivers/gpu/drm/vmwgfx/vmwgfx_drv.c') diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index 66e00b404ec3..3673157527ff 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -73,7 +73,6 @@ static void ttm_bo_del_from_lru(struct ttm_buffer_object *bo) { struct ttm_device *bdev = bo->bdev; - list_del_init(&bo->swap); list_del_init(&bo->lru); if (bdev->funcs->del_from_lru_notify) @@ -105,16 +104,6 @@ void ttm_bo_move_to_lru_tail(struct ttm_buffer_object *bo, man = ttm_manager_type(bdev, mem->mem_type); list_move_tail(&bo->lru, &man->lru[bo->priority]); - if (man->use_tt && bo->ttm && - !(bo->ttm->page_flags & (TTM_PAGE_FLAG_SG | - TTM_PAGE_FLAG_SWAPPED))) { - struct list_head *swap; - - swap = &ttm_glob.swap_lru[bo->priority]; - list_move_tail(&bo->swap, swap); - } else { - list_del_init(&bo->swap); - } if (bdev->funcs->del_from_lru_notify) bdev->funcs->del_from_lru_notify(bo); @@ -129,9 +118,6 @@ void ttm_bo_move_to_lru_tail(struct ttm_buffer_object *bo, ttm_bo_bulk_move_set_pos(&bulk->vram[bo->priority], bo); break; } - if (bo->ttm && !(bo->ttm->page_flags & - (TTM_PAGE_FLAG_SG | TTM_PAGE_FLAG_SWAPPED))) - ttm_bo_bulk_move_set_pos(&bulk->swap[bo->priority], bo); } } EXPORT_SYMBOL(ttm_bo_move_to_lru_tail); @@ -169,20 +155,6 @@ void ttm_bo_bulk_move_lru_tail(struct ttm_lru_bulk_move *bulk) list_bulk_move_tail(&man->lru[i], &pos->first->lru, &pos->last->lru); } - - for (i = 0; i < TTM_MAX_BO_PRIORITY; ++i) { - struct ttm_lru_bulk_move_pos *pos = &bulk->swap[i]; - struct list_head *lru; - - if (!pos->first) - continue; - - dma_resv_assert_held(pos->first->base.resv); - dma_resv_assert_held(pos->last->base.resv); - - lru = &ttm_glob.swap_lru[i]; - list_bulk_move_tail(lru, &pos->first->swap, &pos->last->swap); - } } EXPORT_SYMBOL(ttm_bo_bulk_move_lru_tail); @@ -1065,7 +1037,6 @@ int ttm_bo_init_reserved(struct ttm_device *bdev, kref_init(&bo->kref); INIT_LIST_HEAD(&bo->lru); INIT_LIST_HEAD(&bo->ddestroy); - INIT_LIST_HEAD(&bo->swap); bo->bdev = bdev; bo->type = type; bo->mem.mem_type = TTM_PL_SYSTEM; diff --git a/drivers/gpu/drm/ttm/ttm_bo_util.c b/drivers/gpu/drm/ttm/ttm_bo_util.c index 031e5819fec4..a2a17c84ceb3 100644 --- a/drivers/gpu/drm/ttm/ttm_bo_util.c +++ b/drivers/gpu/drm/ttm/ttm_bo_util.c @@ -303,7 +303,6 @@ static int ttm_buffer_object_transfer(struct ttm_buffer_object *bo, atomic_inc(&ttm_glob.bo_count); INIT_LIST_HEAD(&fbo->base.ddestroy); INIT_LIST_HEAD(&fbo->base.lru); - INIT_LIST_HEAD(&fbo->base.swap); fbo->base.moving = NULL; drm_vma_node_reset(&fbo->base.base.vma_node); diff --git a/drivers/gpu/drm/ttm/ttm_device.c b/drivers/gpu/drm/ttm/ttm_device.c index b1424189fdfb..b9e3bfb60cd2 100644 --- a/drivers/gpu/drm/ttm/ttm_device.c +++ b/drivers/gpu/drm/ttm/ttm_device.c @@ -67,7 +67,6 @@ static int ttm_global_init(void) unsigned long num_pages; struct sysinfo si; int ret = 0; - unsigned i; mutex_lock(&ttm_global_mutex); if (++ttm_glob_use_count > 1) @@ -90,8 +89,6 @@ static int ttm_global_init(void) goto out; } - for (i = 0; i < TTM_MAX_BO_PRIORITY; ++i) - INIT_LIST_HEAD(&glob->swap_lru[i]); INIT_LIST_HEAD(&glob->device_list); atomic_set(&glob->bo_count, 0); @@ -109,27 +106,60 @@ out: int ttm_global_swapout(struct ttm_operation_ctx *ctx, gfp_t gfp_flags) { struct ttm_global *glob = &ttm_glob; + struct ttm_device *bdev; + int ret = -EBUSY; + + mutex_lock(&ttm_global_mutex); + list_for_each_entry(bdev, &glob->device_list, device_list) { + ret = ttm_device_swapout(bdev, ctx, gfp_flags); + if (ret > 0) { + list_move_tail(&bdev->device_list, &glob->device_list); + break; + } + } + mutex_unlock(&ttm_global_mutex); + return ret; +} +EXPORT_SYMBOL(ttm_global_swapout); + +int ttm_device_swapout(struct ttm_device *bdev, struct ttm_operation_ctx *ctx, + gfp_t gfp_flags) +{ + struct ttm_global *glob = &ttm_glob; + struct ttm_resource_manager *man; struct ttm_buffer_object *bo; - unsigned i; + unsigned i, j; int ret; spin_lock(&glob->lru_lock); - for (i = 0; i < TTM_MAX_BO_PRIORITY; ++i) { - list_for_each_entry(bo, &glob->swap_lru[i], swap) { - uint32_t num_pages = bo->ttm->num_pages; - - ret = ttm_bo_swapout(bo, ctx, gfp_flags); - /* ttm_bo_swapout has dropped the lru_lock */ - if (!ret) - return num_pages; - if (ret != -EBUSY) - return ret; + for (i = TTM_PL_SYSTEM; i < TTM_NUM_MEM_TYPES; ++i) { + man = ttm_manager_type(bdev, i); + if (!man || !man->use_tt) + continue; + + for (j = 0; j < TTM_MAX_BO_PRIORITY; ++j) { + list_for_each_entry(bo, &man->lru[j], lru) { + uint32_t num_pages; + + if (!bo->ttm || + bo->ttm->page_flags & TTM_PAGE_FLAG_SG || + bo->ttm->page_flags & TTM_PAGE_FLAG_SWAPPED) + continue; + + num_pages = bo->ttm->num_pages; + ret = ttm_bo_swapout(bo, ctx, gfp_flags); + /* ttm_bo_swapout has dropped the lru_lock */ + if (!ret) + return num_pages; + if (ret != -EBUSY) + return ret; + } } } spin_unlock(&glob->lru_lock); return 0; } -EXPORT_SYMBOL(ttm_global_swapout); +EXPORT_SYMBOL(ttm_device_swapout); static void ttm_init_sysman(struct ttm_device *bdev) { diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c index b991422e156c..4e41d8221f06 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.c @@ -1371,7 +1371,7 @@ static int vmw_pm_freeze(struct device *kdev) vmw_execbuf_release_pinned_bo(dev_priv); vmw_resource_evict_all(dev_priv); vmw_release_device_early(dev_priv); - while (ttm_global_swapout(&ctx, GFP_KERNEL) > 0); + while (ttm_device_swapout(&dev_priv->bdev, &ctx, GFP_KERNEL) > 0); if (dev_priv->enable_fb) vmw_fifo_resource_dec(dev_priv); if (atomic_read(&dev_priv->num_fifo_resources) != 0) { diff --git a/include/drm/ttm/ttm_bo_api.h b/include/drm/ttm/ttm_bo_api.h index 5044ac330858..3587f660e8f4 100644 --- a/include/drm/ttm/ttm_bo_api.h +++ b/include/drm/ttm/ttm_bo_api.h @@ -144,7 +144,6 @@ struct ttm_buffer_object { struct list_head lru; struct list_head ddestroy; - struct list_head swap; /** * Members protected by a bo reservation. diff --git a/include/drm/ttm/ttm_bo_driver.h b/include/drm/ttm/ttm_bo_driver.h index 8959c0075cfd..d007feef7676 100644 --- a/include/drm/ttm/ttm_bo_driver.h +++ b/include/drm/ttm/ttm_bo_driver.h @@ -69,7 +69,6 @@ struct ttm_lru_bulk_move_pos { struct ttm_lru_bulk_move { struct ttm_lru_bulk_move_pos tt[TTM_MAX_BO_PRIORITY]; struct ttm_lru_bulk_move_pos vram[TTM_MAX_BO_PRIORITY]; - struct ttm_lru_bulk_move_pos swap[TTM_MAX_BO_PRIORITY]; }; /* diff --git a/include/drm/ttm/ttm_device.h b/include/drm/ttm/ttm_device.h index ee6bba7fab6f..93f3b59755ac 100644 --- a/include/drm/ttm/ttm_device.h +++ b/include/drm/ttm/ttm_device.h @@ -63,11 +63,6 @@ extern struct ttm_global { */ struct list_head device_list; - /** - * Protected by the lru_lock. - */ - struct list_head swap_lru[TTM_MAX_BO_PRIORITY]; - /** * Internal protection. */ @@ -298,6 +293,8 @@ struct ttm_device { }; int ttm_global_swapout(struct ttm_operation_ctx *ctx, gfp_t gfp_flags); +int ttm_device_swapout(struct ttm_device *bdev, struct ttm_operation_ctx *ctx, + gfp_t gfp_flags); static inline struct ttm_resource_manager * ttm_manager_type(struct ttm_device *bdev, int mem_type) -- cgit v1.2.3