drm/vmwgfx: Stop using raw ttm_buffer_object's
authorZack Rusin <zackr@vmware.com>
Tue, 31 Jan 2023 03:35:42 +0000 (22:35 -0500)
committerZack Rusin <zackr@vmware.com>
Tue, 14 Feb 2023 03:37:55 +0000 (22:37 -0500)
Various bits of the driver used raw ttm_buffer_object instead of the
driver specific vmw_bo object. All those places used to duplicate
the mapped bo caching policy of vmw_bo.

Instead of duplicating all of that code and special casing various
functions to work both with vmw_bo and raw ttm_buffer_object's unify
the buffer object handling code.

As part of that work fix the naming of bo's, e.g. insted of generic
backup use 'guest_memory' because that's what it really is.

All of it makes the driver easier to maintain and the code easier to
read. Saves 100+ loc as well.

Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Martin Krastev <krastevm@vmware.com>
Reviewed-by: Maaz Mombasawala <mombasawalam@vmware.com>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20230131033542.953249-9-zack@kde.org
27 files changed:
drivers/gpu/drm/vmwgfx/vmwgfx_bo.c
drivers/gpu/drm/vmwgfx/vmwgfx_bo.h
drivers/gpu/drm/vmwgfx/vmwgfx_cmd.c
drivers/gpu/drm/vmwgfx/vmwgfx_cmdbuf.c
drivers/gpu/drm/vmwgfx/vmwgfx_context.c
drivers/gpu/drm/vmwgfx/vmwgfx_cotable.c
drivers/gpu/drm/vmwgfx/vmwgfx_drv.c
drivers/gpu/drm/vmwgfx/vmwgfx_drv.h
drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c
drivers/gpu/drm/vmwgfx/vmwgfx_gem.c
drivers/gpu/drm/vmwgfx/vmwgfx_kms.c
drivers/gpu/drm/vmwgfx/vmwgfx_kms.h
drivers/gpu/drm/vmwgfx/vmwgfx_ldu.c
drivers/gpu/drm/vmwgfx/vmwgfx_mob.c
drivers/gpu/drm/vmwgfx/vmwgfx_overlay.c
drivers/gpu/drm/vmwgfx/vmwgfx_page_dirty.c
drivers/gpu/drm/vmwgfx/vmwgfx_resource.c
drivers/gpu/drm/vmwgfx/vmwgfx_resource_priv.h
drivers/gpu/drm/vmwgfx/vmwgfx_scrn.c
drivers/gpu/drm/vmwgfx/vmwgfx_shader.c
drivers/gpu/drm/vmwgfx/vmwgfx_so.c
drivers/gpu/drm/vmwgfx/vmwgfx_stdu.c
drivers/gpu/drm/vmwgfx/vmwgfx_streamoutput.c
drivers/gpu/drm/vmwgfx/vmwgfx_surface.c
drivers/gpu/drm/vmwgfx/vmwgfx_ttm_buffer.c
drivers/gpu/drm/vmwgfx/vmwgfx_va.c
drivers/gpu/drm/vmwgfx/vmwgfx_validation.c

index dbee64cd40fe38ff02efff73feb5ca335bdccaa9..ee8f87dcdf21d13ae292b72c642a88e35722cb3d 100644 (file)
 
 #include <drm/ttm/ttm_placement.h>
 
+static void vmw_bo_release(struct vmw_bo *vbo)
+{
+       vmw_bo_unmap(vbo);
+       drm_gem_object_release(&vbo->tbo.base);
+}
+
 /**
  * vmw_bo_free - vmw_bo destructor
  *
@@ -43,26 +49,10 @@ static void vmw_bo_free(struct ttm_buffer_object *bo)
 
        WARN_ON(vbo->dirty);
        WARN_ON(!RB_EMPTY_ROOT(&vbo->res_tree));
-       vmw_bo_unmap(vbo);
-       drm_gem_object_release(&bo->base);
+       vmw_bo_release(vbo);
        kfree(vbo);
 }
 
-/**
- * bo_is_vmw - check if the buffer object is a &vmw_bo
- * @bo: ttm buffer object to be checked
- *
- * Uses destroy function associated with the object to determine if this is
- * a &vmw_bo.
- *
- * Returns:
- * true if the object is of &vmw_bo type, false if not.
- */
-static bool bo_is_vmw(struct ttm_buffer_object *bo)
-{
-       return bo->destroy == &vmw_bo_free;
-}
-
 /**
  * vmw_bo_pin_in_placement - Validate a buffer to placement.
  *
@@ -79,7 +69,7 @@ static int vmw_bo_pin_in_placement(struct vmw_private *dev_priv,
                                   bool interruptible)
 {
        struct ttm_operation_ctx ctx = {interruptible, false };
-       struct ttm_buffer_object *bo = &buf->base;
+       struct ttm_buffer_object *bo = &buf->tbo;
        int ret;
 
        vmw_execbuf_release_pinned_bo(dev_priv);
@@ -115,7 +105,7 @@ int vmw_bo_pin_in_vram_or_gmr(struct vmw_private *dev_priv,
                              bool interruptible)
 {
        struct ttm_operation_ctx ctx = {interruptible, false };
-       struct ttm_buffer_object *bo = &buf->base;
+       struct ttm_buffer_object *bo = &buf->tbo;
        int ret;
 
        vmw_execbuf_release_pinned_bo(dev_priv);
@@ -184,7 +174,7 @@ int vmw_bo_pin_in_start_of_vram(struct vmw_private *dev_priv,
                                bool interruptible)
 {
        struct ttm_operation_ctx ctx = {interruptible, false };
-       struct ttm_buffer_object *bo = &buf->base;
+       struct ttm_buffer_object *bo = &buf->tbo;
        int ret = 0;
 
        vmw_execbuf_release_pinned_bo(dev_priv);
@@ -200,7 +190,7 @@ int vmw_bo_pin_in_start_of_vram(struct vmw_private *dev_priv,
        if (bo->resource->mem_type == TTM_PL_VRAM &&
            bo->resource->start < PFN_UP(bo->resource->size) &&
            bo->resource->start > 0 &&
-           buf->base.pin_count == 0) {
+           buf->tbo.pin_count == 0) {
                ctx.interruptible = false;
                vmw_bo_placement_set(buf,
                                     VMW_BO_DOMAIN_SYS,
@@ -241,7 +231,7 @@ int vmw_bo_unpin(struct vmw_private *dev_priv,
                 struct vmw_bo *buf,
                 bool interruptible)
 {
-       struct ttm_buffer_object *bo = &buf->base;
+       struct ttm_buffer_object *bo = &buf->tbo;
        int ret;
 
        ret = ttm_bo_reserve(bo, interruptible, false, NULL);
@@ -288,7 +278,7 @@ void vmw_bo_pin_reserved(struct vmw_bo *vbo, bool pin)
        struct ttm_operation_ctx ctx = { false, true };
        struct ttm_place pl;
        struct ttm_placement placement;
-       struct ttm_buffer_object *bo = &vbo->base;
+       struct ttm_buffer_object *bo = &vbo->tbo;
        uint32_t old_mem_type = bo->resource->mem_type;
        int ret;
 
@@ -333,7 +323,7 @@ void vmw_bo_pin_reserved(struct vmw_bo *vbo, bool pin)
  */
 void *vmw_bo_map_and_cache(struct vmw_bo *vbo)
 {
-       struct ttm_buffer_object *bo = &vbo->base;
+       struct ttm_buffer_object *bo = &vbo->tbo;
        bool not_used;
        void *virtual;
        int ret;
@@ -364,64 +354,58 @@ void vmw_bo_unmap(struct vmw_bo *vbo)
                return;
 
        ttm_bo_kunmap(&vbo->map);
+       vbo->map.bo = NULL;
 }
 
-/* default destructor */
-static void vmw_bo_default_destroy(struct ttm_buffer_object *bo)
-{
-       kfree(bo);
-}
 
 /**
- * vmw_bo_create_kernel - Create a pinned BO for internal kernel use.
+ * vmw_bo_init - Initialize a vmw buffer object
  *
  * @dev_priv: Pointer to the device private struct
- * @size: size of the BO we need
- * @placement: where to put it
- * @p_bo: resulting BO
+ * @vmw_bo: Buffer object to initialize
+ * @params: Parameters used to initialize the buffer object
+ * @destroy: The function used to delete the buffer object
+ * Returns: Zero on success, negative error code on error.
  *
- * Creates and pin a simple BO for in kernel use.
  */
-int vmw_bo_create_kernel(struct vmw_private *dev_priv, unsigned long size,
-                        struct ttm_placement *placement,
-                        struct ttm_buffer_object **p_bo)
+static int vmw_bo_init(struct vmw_private *dev_priv,
+                      struct vmw_bo *vmw_bo,
+                      struct vmw_bo_params *params,
+                      void (*destroy)(struct ttm_buffer_object *))
 {
        struct ttm_operation_ctx ctx = {
-               .interruptible = false,
+               .interruptible = params->bo_type != ttm_bo_type_kernel,
                .no_wait_gpu = false
        };
-       struct ttm_buffer_object *bo;
+       struct ttm_device *bdev = &dev_priv->bdev;
        struct drm_device *vdev = &dev_priv->drm;
        int ret;
 
-       bo = kzalloc(sizeof(*bo), GFP_KERNEL);
-       if (unlikely(!bo))
-               return -ENOMEM;
+       memset(vmw_bo, 0, sizeof(*vmw_bo));
 
-       size = ALIGN(size, PAGE_SIZE);
+       BUILD_BUG_ON(TTM_MAX_BO_PRIORITY <= 3);
+       vmw_bo->tbo.priority = 3;
+       vmw_bo->res_tree = RB_ROOT;
 
-       drm_gem_private_object_init(vdev, &bo->base, size);
+       params->size = ALIGN(params->size, PAGE_SIZE);
+       drm_gem_private_object_init(vdev, &vmw_bo->tbo.base, params->size);
 
-       ret = ttm_bo_init_reserved(&dev_priv->bdev, bo, ttm_bo_type_kernel,
-                                  placement, 0, &ctx, NULL, NULL,
-                                  vmw_bo_default_destroy);
+       vmw_bo_placement_set(vmw_bo, params->domain, params->busy_domain);
+       ret = ttm_bo_init_reserved(bdev, &vmw_bo->tbo, params->bo_type,
+                                  &vmw_bo->placement, 0, &ctx, NULL,
+                                  NULL, destroy);
        if (unlikely(ret))
-               goto error_free;
+               return ret;
 
-       ttm_bo_pin(bo);
-       ttm_bo_unreserve(bo);
-       *p_bo = bo;
+       if (params->pin)
+               ttm_bo_pin(&vmw_bo->tbo);
+       ttm_bo_unreserve(&vmw_bo->tbo);
 
        return 0;
-
-error_free:
-       kfree(bo);
-       return ret;
 }
 
 int vmw_bo_create(struct vmw_private *vmw,
-                 size_t size, u32 domain, u32 busy_domain,
-                 bool interruptible, bool pin,
+                 struct vmw_bo_params *params,
                  struct vmw_bo **p_bo)
 {
        int ret;
@@ -432,9 +416,7 @@ int vmw_bo_create(struct vmw_private *vmw,
                return -ENOMEM;
        }
 
-       ret = vmw_bo_init(vmw, *p_bo, size,
-                         domain, busy_domain,
-                         interruptible, pin);
+       ret = vmw_bo_init(vmw, *p_bo, params, vmw_bo_free);
        if (unlikely(ret != 0))
                goto out_error;
 
@@ -445,57 +427,6 @@ out_error:
        return ret;
 }
 
-/**
- * vmw_bo_init - Initialize a vmw buffer object
- *
- * @dev_priv: Pointer to the device private struct
- * @vmw_bo: Pointer to the struct vmw_bo to initialize.
- * @size: Buffer object size in bytes.
- * @domain: Domain to put the bo in.
- * @busy_domain: Domain to put the bo if busy.
- * @interruptible: Whether waits should be performed interruptible.
- * @pin: If the BO should be created pinned at a fixed location.
- * Returns: Zero on success, negative error code on error.
- *
- * Note that on error, the code will free the buffer object.
- */
-int vmw_bo_init(struct vmw_private *dev_priv,
-               struct vmw_bo *vmw_bo,
-               size_t size,
-               u32 domain,
-               u32 busy_domain,
-               bool interruptible, bool pin)
-{
-       struct ttm_operation_ctx ctx = {
-               .interruptible = interruptible,
-               .no_wait_gpu = false
-       };
-       struct ttm_device *bdev = &dev_priv->bdev;
-       struct drm_device *vdev = &dev_priv->drm;
-       int ret;
-
-       memset(vmw_bo, 0, sizeof(*vmw_bo));
-       BUILD_BUG_ON(TTM_MAX_BO_PRIORITY <= 3);
-       vmw_bo->base.priority = 3;
-       vmw_bo->res_tree = RB_ROOT;
-
-       size = ALIGN(size, PAGE_SIZE);
-       drm_gem_private_object_init(vdev, &vmw_bo->base.base, size);
-
-       vmw_bo_placement_set(vmw_bo, domain, busy_domain);
-       ret = ttm_bo_init_reserved(bdev, &vmw_bo->base, ttm_bo_type_device,
-                                  &vmw_bo->placement, 0, &ctx, NULL, NULL, vmw_bo_free);
-       if (unlikely(ret)) {
-               return ret;
-       }
-
-       if (pin)
-               ttm_bo_pin(&vmw_bo->base);
-       ttm_bo_unreserve(&vmw_bo->base);
-
-       return 0;
-}
-
 /**
  * vmw_user_bo_synccpu_grab - Grab a struct vmw_bo for cpu
  * access, idling previous GPU operations on the buffer and optionally
@@ -514,7 +445,7 @@ static int vmw_user_bo_synccpu_grab(struct vmw_bo *vmw_bo,
                                    uint32_t flags)
 {
        bool nonblock = !!(flags & drm_vmw_synccpu_dontblock);
-       struct ttm_buffer_object *bo = &vmw_bo->base;
+       struct ttm_buffer_object *bo = &vmw_bo->tbo;
        int ret;
 
        if (flags & drm_vmw_synccpu_allow_cs) {
@@ -564,7 +495,7 @@ static int vmw_user_bo_synccpu_release(struct drm_file *filp,
                if (!(flags & drm_vmw_synccpu_allow_cs)) {
                        atomic_dec(&vmw_bo->cpu_writers);
                }
-               ttm_bo_put(&vmw_bo->base);
+               ttm_bo_put(&vmw_bo->tbo);
        }
 
        return ret;
@@ -650,8 +581,7 @@ int vmw_bo_unref_ioctl(struct drm_device *dev, void *data,
        struct drm_vmw_unref_dmabuf_arg *arg =
            (struct drm_vmw_unref_dmabuf_arg *)data;
 
-       drm_gem_handle_delete(file_priv, arg->handle);
-       return 0;
+       return drm_gem_handle_delete(file_priv, arg->handle);
 }
 
 
@@ -667,7 +597,7 @@ int vmw_bo_unref_ioctl(struct drm_device *dev, void *data,
  * The vmw buffer object pointer will be refcounted.
  */
 int vmw_user_bo_lookup(struct drm_file *filp,
-                      uint32_t handle,
+                      u32 handle,
                       struct vmw_bo **out)
 {
        struct drm_gem_object *gobj;
@@ -680,7 +610,7 @@ int vmw_user_bo_lookup(struct drm_file *filp,
        }
 
        *out = to_vmw_bo(gobj);
-       ttm_bo_get(&(*out)->base);
+       ttm_bo_get(&(*out)->tbo);
        drm_gem_object_put(gobj);
 
        return 0;
@@ -702,8 +632,7 @@ void vmw_bo_fence_single(struct ttm_buffer_object *bo,
                         struct vmw_fence_obj *fence)
 {
        struct ttm_device *bdev = bo->bdev;
-       struct vmw_private *dev_priv =
-               container_of(bdev, struct vmw_private, bdev);
+       struct vmw_private *dev_priv = vmw_priv_from_ttm(bdev);
        int ret;
 
        if (fence == NULL)
@@ -773,10 +702,6 @@ int vmw_dumb_create(struct drm_file *file_priv,
  */
 void vmw_bo_swap_notify(struct ttm_buffer_object *bo)
 {
-       /* Is @bo embedded in a struct vmw_bo? */
-       if (!bo_is_vmw(bo))
-               return;
-
        /* Kill any cached kernel maps before swapout */
        vmw_bo_unmap(to_vmw_bo(&bo->base));
 }
@@ -795,13 +720,7 @@ void vmw_bo_swap_notify(struct ttm_buffer_object *bo)
 void vmw_bo_move_notify(struct ttm_buffer_object *bo,
                        struct ttm_resource *mem)
 {
-       struct vmw_bo *vbo;
-
-       /* Make sure @bo is embedded in a struct vmw_bo? */
-       if (!bo_is_vmw(bo))
-               return;
-
-       vbo = container_of(bo, struct vmw_bo, base);
+       struct vmw_bo *vbo = to_vmw_bo(&bo->base);
 
        /*
         * Kill any cached kernel maps before move to or from VRAM.
@@ -849,7 +768,6 @@ set_placement_list(struct ttm_place *pl, u32 domain)
                pl[n].lpfn = 0;
                n++;
        }
-       WARN_ON((domain & VMW_BO_DOMAIN_WAITABLE_SYS) != 0);
        if (domain & VMW_BO_DOMAIN_WAITABLE_SYS) {
                pl[n].mem_type = VMW_PL_SYSTEM;
                pl[n].flags = 0;
@@ -878,9 +796,8 @@ set_placement_list(struct ttm_place *pl, u32 domain)
 
 void vmw_bo_placement_set(struct vmw_bo *bo, u32 domain, u32 busy_domain)
 {
-       struct ttm_device *bdev = bo->base.bdev;
-       struct vmw_private *vmw =
-               container_of(bdev, struct vmw_private, bdev);
+       struct ttm_device *bdev = bo->tbo.bdev;
+       struct vmw_private *vmw = vmw_priv_from_ttm(bdev);
        struct ttm_placement *pl = &bo->placement;
        bool mem_compatible = false;
        u32 i;
@@ -888,17 +805,17 @@ void vmw_bo_placement_set(struct vmw_bo *bo, u32 domain, u32 busy_domain)
        pl->placement = bo->places;
        pl->num_placement = set_placement_list(bo->places, domain);
 
-       if (drm_debug_enabled(DRM_UT_DRIVER) && bo->base.resource) {
+       if (drm_debug_enabled(DRM_UT_DRIVER) && bo->tbo.resource) {
                for (i = 0; i < pl->num_placement; ++i) {
-                       if (bo->base.resource->mem_type == TTM_PL_SYSTEM ||
-                           bo->base.resource->mem_type == pl->placement[i].mem_type)
+                       if (bo->tbo.resource->mem_type == TTM_PL_SYSTEM ||
+                           bo->tbo.resource->mem_type == pl->placement[i].mem_type)
                                mem_compatible = true;
                }
                if (!mem_compatible)
                        drm_warn(&vmw->drm,
                                 "%s: Incompatible transition from "
                                 "bo->base.resource->mem_type = %u to domain = %u\n",
-                                __func__, bo->base.resource->mem_type, domain);
+                                __func__, bo->tbo.resource->mem_type, domain);
        }
 
        pl->busy_placement = bo->busy_places;
@@ -907,9 +824,8 @@ void vmw_bo_placement_set(struct vmw_bo *bo, u32 domain, u32 busy_domain)
 
 void vmw_bo_placement_set_default_accelerated(struct vmw_bo *bo)
 {
-       struct ttm_device *bdev = bo->base.bdev;
-       struct vmw_private *vmw =
-               container_of(bdev, struct vmw_private, bdev);
+       struct ttm_device *bdev = bo->tbo.bdev;
+       struct vmw_private *vmw = vmw_priv_from_ttm(bdev);
        u32 domain = VMW_BO_DOMAIN_GMR | VMW_BO_DOMAIN_VRAM;
 
        if (vmw->has_mob)
index f146e8e80515b394b1e80c1a239d071dbc2386d0..50a836e70994934a7dd8dd2c1c2b36a55cc64f4c 100644 (file)
@@ -49,54 +49,54 @@ enum vmw_bo_domain {
        VMW_BO_DOMAIN_MOB           = BIT(4),
 };
 
+struct vmw_bo_params {
+       u32 domain;
+       u32 busy_domain;
+       enum ttm_bo_type bo_type;
+       size_t size;
+       bool pin;
+};
+
 /**
  * struct vmw_bo - TTM buffer object with vmwgfx additions
- * @base: The TTM buffer object
+ * @tbo: The TTM buffer object
+ * @placement: The preferred placement for this buffer object
+ * @places: The chosen places for the preferred placement.
+ * @busy_places: Chosen busy places for the preferred placement
+ * @map: Kmap object for semi-persistent mappings
  * @res_tree: RB tree of resources using this buffer object as a backing MOB
+ * @res_prios: Eviction priority counts for attached resources
  * @cpu_writers: Number of synccpu write grabs. Protected by reservation when
  * increased. May be decreased without reservation.
  * @dx_query_ctx: DX context if this buffer object is used as a DX query MOB
- * @map: Kmap object for semi-persistent mappings
- * @res_prios: Eviction priority counts for attached resources
  * @dirty: structure for user-space dirty-tracking
  */
 struct vmw_bo {
-       struct ttm_buffer_object base;
+       struct ttm_buffer_object tbo;
 
        struct ttm_placement placement;
        struct ttm_place places[5];
        struct ttm_place busy_places[5];
 
+       /* Protected by reservation */
+       struct ttm_bo_kmap_obj map;
+
        struct rb_root res_tree;
+       u32 res_prios[TTM_MAX_BO_PRIORITY];
 
        atomic_t cpu_writers;
        /* Not ref-counted.  Protected by binding_mutex */
        struct vmw_resource *dx_query_ctx;
-       /* Protected by reservation */
-       struct ttm_bo_kmap_obj map;
-       u32 res_prios[TTM_MAX_BO_PRIORITY];
        struct vmw_bo_dirty *dirty;
 };
 
 void vmw_bo_placement_set(struct vmw_bo *bo, u32 domain, u32 busy_domain);
 void vmw_bo_placement_set_default_accelerated(struct vmw_bo *bo);
 
-int vmw_bo_create_kernel(struct vmw_private *dev_priv,
-                        unsigned long size,
-                        struct ttm_placement *placement,
-                        struct ttm_buffer_object **p_bo);
 int vmw_bo_create(struct vmw_private *dev_priv,
-                 size_t size,
-                 u32 domain,
-                 u32 busy_domain,
-                 bool interruptible, bool pin,
+                 struct vmw_bo_params *params,
                  struct vmw_bo **p_bo);
-int vmw_bo_init(struct vmw_private *dev_priv,
-               struct vmw_bo *vmw_bo,
-               size_t size,
-               u32 domain,
-               u32 busy_domain,
-               bool interruptible, bool pin);
+
 int vmw_bo_unref_ioctl(struct drm_device *dev, void *data,
                       struct drm_file *file_priv);
 
@@ -118,9 +118,6 @@ void vmw_bo_get_guest_ptr(const struct ttm_buffer_object *buf,
                          SVGAGuestPtr *ptr);
 int vmw_user_bo_synccpu_ioctl(struct drm_device *dev, void *data,
                              struct drm_file *file_priv);
-int vmw_user_bo_lookup(struct drm_file *filp,
-                      uint32_t handle,
-                      struct vmw_bo **out);
 void vmw_bo_fence_single(struct ttm_buffer_object *bo,
                         struct vmw_fence_obj *fence);
 
@@ -131,6 +128,9 @@ void vmw_bo_move_notify(struct ttm_buffer_object *bo,
                        struct ttm_resource *mem);
 void vmw_bo_swap_notify(struct ttm_buffer_object *bo);
 
+int vmw_user_bo_lookup(struct drm_file *filp,
+                      u32 handle,
+                      struct vmw_bo **out);
 /**
  * vmw_bo_adjust_prio - Adjust the buffer object eviction priority
  * according to attached resources
@@ -142,12 +142,12 @@ static inline void vmw_bo_prio_adjust(struct vmw_bo *vbo)
 
        while (i--) {
                if (vbo->res_prios[i]) {
-                       vbo->base.priority = i;
+                       vbo->tbo.priority = i;
                        return;
                }
        }
 
-       vbo->base.priority = 3;
+       vbo->tbo.priority = 3;
 }
 
 /**
@@ -166,7 +166,7 @@ static inline void vmw_bo_prio_add(struct vmw_bo *vbo, int prio)
 }
 
 /**
- * vmw_bo_prio_del - Notify a buffer object of a resource with a certain
+ * vmw_bo_used_prio_del - Notify a buffer object of a resource with a certain
  * priority being removed
  * @vbo: The struct vmw_bo
  * @prio: The resource priority
@@ -186,18 +186,18 @@ static inline void vmw_bo_unreference(struct vmw_bo **buf)
 
        *buf = NULL;
        if (tmp_buf)
-               ttm_bo_put(&tmp_buf->base);
+               ttm_bo_put(&tmp_buf->tbo);
 }
 
 static inline struct vmw_bo *vmw_bo_reference(struct vmw_bo *buf)
 {
-       ttm_bo_get(&buf->base);
+       ttm_bo_get(&buf->tbo);
        return buf;
 }
 
 static inline struct vmw_bo *to_vmw_bo(struct drm_gem_object *gobj)
 {
-       return container_of((gobj), struct vmw_bo, base.base);
+       return container_of((gobj), struct vmw_bo, tbo.base);
 }
 
 #endif // VMWGFX_BO_H
index b1e7810032d3c9dfa3864bc07c90901deb1d689b..195ff8792e5aba71399d9b4ca6fa3ceab4ad8add 100644 (file)
@@ -567,7 +567,7 @@ static int vmw_cmd_emit_dummy_legacy_query(struct vmw_private *dev_priv,
         * without writing to the query result structure.
         */
 
-       struct ttm_buffer_object *bo = &dev_priv->dummy_query_bo->base;
+       struct ttm_buffer_object *bo = &dev_priv->dummy_query_bo->tbo;
        struct {
                SVGA3dCmdHeader header;
                SVGA3dCmdWaitForQuery body;
@@ -613,7 +613,7 @@ static int vmw_cmd_emit_dummy_gb_query(struct vmw_private *dev_priv,
         * without writing to the query result structure.
         */
 
-       struct ttm_buffer_object *bo = &dev_priv->dummy_query_bo->base;
+       struct ttm_buffer_object *bo = &dev_priv->dummy_query_bo->tbo;
        struct {
                SVGA3dCmdHeader header;
                SVGA3dCmdWaitForGBQuery body;
index 11ff367692ce5c57d9a2603b672450c8f7f1a5c6..94e8982f5616921d9d8f5ca1f9711730aca937a2 100644 (file)
@@ -80,7 +80,6 @@ struct vmw_cmdbuf_context {
  * frees are protected by @lock.
  * @cmd_space: Buffer object for the command buffer space, unless we were
  * able to make a contigous coherent DMA memory allocation, @handle. Immutable.
- * @map_obj: Mapping state for @cmd_space. Immutable.
  * @map: Pointer to command buffer space. May be a mapped buffer object or
  * a contigous coherent DMA memory allocation. Immutable.
  * @cur: Command buffer for small kernel command submissions. Protected by
@@ -117,8 +116,7 @@ struct vmw_cmdbuf_man {
        struct vmw_cmdbuf_context ctx[SVGA_CB_CONTEXT_MAX];
        struct list_head error;
        struct drm_mm mm;
-       struct ttm_buffer_object *cmd_space;
-       struct ttm_bo_kmap_obj map_obj;
+       struct vmw_bo *cmd_space;
        u8 *map;
        struct vmw_cmdbuf_header *cur;
        size_t cur_pos;
@@ -889,7 +887,7 @@ static int vmw_cmdbuf_space_pool(struct vmw_cmdbuf_man *man,
        header->cmd = man->map + offset;
        if (man->using_mob) {
                cb_hdr->flags = SVGA_CB_FLAG_MOB;
-               cb_hdr->ptr.mob.mobid = man->cmd_space->resource->start;
+               cb_hdr->ptr.mob.mobid = man->cmd_space->tbo.resource->start;
                cb_hdr->ptr.mob.mobOffset = offset;
        } else {
                cb_hdr->ptr.pa = (u64)man->handle + (u64)offset;
@@ -1222,7 +1220,6 @@ static int vmw_cmdbuf_startstop(struct vmw_cmdbuf_man *man, u32 context,
 int vmw_cmdbuf_set_pool_size(struct vmw_cmdbuf_man *man, size_t size)
 {
        struct vmw_private *dev_priv = man->dev_priv;
-       bool dummy;
        int ret;
 
        if (man->has_pool)
@@ -1235,6 +1232,13 @@ int vmw_cmdbuf_set_pool_size(struct vmw_cmdbuf_man *man, size_t size)
        if (man->map) {
                man->using_mob = false;
        } else {
+               struct vmw_bo_params bo_params = {
+                       .domain = VMW_BO_DOMAIN_MOB,
+                       .busy_domain = VMW_BO_DOMAIN_MOB,
+                       .bo_type = ttm_bo_type_kernel,
+                       .size = size,
+                       .pin = true
+               };
                /*
                 * DMA memory failed. If we can have command buffers in a
                 * MOB, try to use that instead. Note that this will
@@ -1245,19 +1249,12 @@ int vmw_cmdbuf_set_pool_size(struct vmw_cmdbuf_man *man, size_t size)
                    !dev_priv->has_mob)
                        return -ENOMEM;
 
-               ret = vmw_bo_create_kernel(dev_priv, size,
-                                          &vmw_mob_placement,
-                                          &man->cmd_space);
+               ret = vmw_bo_create(dev_priv, &bo_params, &man->cmd_space);
                if (ret)
                        return ret;
 
-               man->using_mob = true;
-               ret = ttm_bo_kmap(man->cmd_space, 0, size >> PAGE_SHIFT,
-                                 &man->map_obj);
-               if (ret)
-                       goto out_no_map;
-
-               man->map = ttm_kmap_obj_virtual(&man->map_obj, &dummy);
+               man->map = vmw_bo_map_and_cache(man->cmd_space);
+               man->using_mob = man->map;
        }
 
        man->size = size;
@@ -1277,14 +1274,6 @@ int vmw_cmdbuf_set_pool_size(struct vmw_cmdbuf_man *man, size_t size)
                 (man->using_mob) ? "MOB" : "DMA");
 
        return 0;
-
-out_no_map:
-       if (man->using_mob) {
-               ttm_bo_put(man->cmd_space);
-               man->cmd_space = NULL;
-       }
-
-       return ret;
 }
 
 /**
@@ -1383,14 +1372,11 @@ void vmw_cmdbuf_remove_pool(struct vmw_cmdbuf_man *man)
        man->has_pool = false;
        man->default_size = VMW_CMDBUF_INLINE_SIZE;
        (void) vmw_cmdbuf_idle(man, false, 10*HZ);
-       if (man->using_mob) {
-               (void) ttm_bo_kunmap(&man->map_obj);
-               ttm_bo_put(man->cmd_space);
-               man->cmd_space = NULL;
-       } else {
+       if (man->using_mob)
+               vmw_bo_unreference(&man->cmd_space);
+       else
                dma_free_coherent(man->dev_priv->drm.dev,
                                  man->size, man->map, man->handle);
-       }
 }
 
 /**
index 84ef5f0a785bbff27357e0772f5384d6cff98f0c..ecc503e42790a784b3c8a7efa7773c164e630f4b 100644 (file)
@@ -73,7 +73,7 @@ const struct vmw_user_resource_conv *user_context_converter =
 
 static const struct vmw_res_func vmw_legacy_context_func = {
        .res_type = vmw_res_context,
-       .needs_backup = false,
+       .needs_guest_memory = false,
        .may_evict = false,
        .type_name = "legacy contexts",
        .domain = VMW_BO_DOMAIN_SYS,
@@ -86,7 +86,7 @@ static const struct vmw_res_func vmw_legacy_context_func = {
 
 static const struct vmw_res_func vmw_gb_context_func = {
        .res_type = vmw_res_context,
-       .needs_backup = true,
+       .needs_guest_memory = true,
        .may_evict = true,
        .prio = 3,
        .dirty_prio = 3,
@@ -101,7 +101,7 @@ static const struct vmw_res_func vmw_gb_context_func = {
 
 static const struct vmw_res_func vmw_dx_context_func = {
        .res_type = vmw_res_dx_context,
-       .needs_backup = true,
+       .needs_guest_memory = true,
        .may_evict = true,
        .prio = 3,
        .dirty_prio = 3,
@@ -186,7 +186,7 @@ static int vmw_gb_context_init(struct vmw_private *dev_priv,
        struct vmw_user_context *uctx =
                container_of(res, struct vmw_user_context, res);
 
-       res->backup_size = (dx ? sizeof(SVGADXContextMobFormat) :
+       res->guest_memory_size = (dx ? sizeof(SVGADXContextMobFormat) :
                                 sizeof(SVGAGBContextData));
        ret = vmw_resource_init(dev_priv, res, true,
                                res_free,
@@ -358,8 +358,8 @@ static int vmw_gb_context_bind(struct vmw_resource *res,
        cmd->header.size = sizeof(cmd->body);
        cmd->body.cid = res->id;
        cmd->body.mobid = bo->resource->start;
-       cmd->body.validContents = res->backup_dirty;
-       res->backup_dirty = false;
+       cmd->body.validContents = res->guest_memory_dirty;
+       res->guest_memory_dirty = false;
        vmw_cmd_commit(dev_priv, sizeof(*cmd));
 
        return 0;
@@ -525,8 +525,8 @@ static int vmw_dx_context_bind(struct vmw_resource *res,
        cmd->header.size = sizeof(cmd->body);
        cmd->body.cid = res->id;
        cmd->body.mobid = bo->resource->start;
-       cmd->body.validContents = res->backup_dirty;
-       res->backup_dirty = false;
+       cmd->body.validContents = res->guest_memory_dirty;
+       res->guest_memory_dirty = false;
        vmw_cmd_commit(dev_priv, sizeof(*cmd));
 
 
index d49db8146df107779ecd137399c9c97f62f7bc31..c0b24d1cacbf0bd3392817d35a4501316455d34c 100644 (file)
@@ -131,7 +131,7 @@ static int vmw_cotable_destroy(struct vmw_resource *res);
 
 static const struct vmw_res_func vmw_cotable_func = {
        .res_type = vmw_res_cotable,
-       .needs_backup = true,
+       .needs_guest_memory = true,
        .may_evict = true,
        .prio = 3,
        .dirty_prio = 3,
@@ -182,7 +182,7 @@ static int vmw_cotable_unscrub(struct vmw_resource *res)
 {
        struct vmw_cotable *vcotbl = vmw_cotable(res);
        struct vmw_private *dev_priv = res->dev_priv;
-       struct ttm_buffer_object *bo = &res->backup->base;
+       struct ttm_buffer_object *bo = &res->guest_memory_bo->tbo;
        struct {
                SVGA3dCmdHeader header;
                SVGA3dCmdDXSetCOTable body;
@@ -230,7 +230,7 @@ static int vmw_cotable_bind(struct vmw_resource *res,
         * take the opportunity to correct the value here so that it's not
         * misused in the future.
         */
-       val_buf->bo = &res->backup->base;
+       val_buf->bo = &res->guest_memory_bo->tbo;
 
        return vmw_cotable_unscrub(res);
 }
@@ -291,7 +291,7 @@ int vmw_cotable_scrub(struct vmw_resource *res, bool readback)
                cmd0->body.cid = vcotbl->ctx->id;
                cmd0->body.type = vcotbl->type;
                cmd1 = (void *) &cmd0[1];
-               vcotbl->size_read_back = res->backup_size;
+               vcotbl->size_read_back = res->guest_memory_size;
        }
        cmd1->header.id = SVGA_3D_CMD_DX_SET_COTABLE;
        cmd1->header.size = sizeof(cmd1->body);
@@ -373,12 +373,12 @@ static int vmw_cotable_readback(struct vmw_resource *res)
                cmd->header.size = sizeof(cmd->body);
                cmd->body.cid = vcotbl->ctx->id;
                cmd->body.type = vcotbl->type;
-               vcotbl->size_read_back = res->backup_size;
+               vcotbl->size_read_back = res->guest_memory_size;
                vmw_cmd_commit(dev_priv, sizeof(*cmd));
        }
 
        (void) vmw_execbuf_fence_commands(NULL, dev_priv, &fence, NULL);
-       vmw_bo_fence_single(&res->backup->base, fence);
+       vmw_bo_fence_single(&res->guest_memory_bo->tbo, fence);
        vmw_fence_obj_unreference(&fence);
 
        return 0;
@@ -401,14 +401,21 @@ static int vmw_cotable_resize(struct vmw_resource *res, size_t new_size)
        struct ttm_operation_ctx ctx = { false, false };
        struct vmw_private *dev_priv = res->dev_priv;
        struct vmw_cotable *vcotbl = vmw_cotable(res);
-       struct vmw_bo *buf, *old_buf = res->backup;
-       struct ttm_buffer_object *bo, *old_bo = &res->backup->base;
-       size_t old_size = res->backup_size;
+       struct vmw_bo *buf, *old_buf = res->guest_memory_bo;
+       struct ttm_buffer_object *bo, *old_bo = &res->guest_memory_bo->tbo;
+       size_t old_size = res->guest_memory_size;
        size_t old_size_read_back = vcotbl->size_read_back;
        size_t cur_size_read_back;
        struct ttm_bo_kmap_obj old_map, new_map;
        int ret;
        size_t i;
+       struct vmw_bo_params bo_params = {
+               .domain = VMW_BO_DOMAIN_MOB,
+               .busy_domain = VMW_BO_DOMAIN_MOB,
+               .bo_type = ttm_bo_type_device,
+               .size = new_size,
+               .pin = true
+       };
 
        MKS_STAT_TIME_DECL(MKSSTAT_KERN_COTABLE_RESIZE);
        MKS_STAT_TIME_PUSH(MKSSTAT_KERN_COTABLE_RESIZE);
@@ -425,15 +432,13 @@ static int vmw_cotable_resize(struct vmw_resource *res, size_t new_size)
         * for the new COTable. Initially pin the buffer object to make sure
         * we can use tryreserve without failure.
         */
-       ret = vmw_bo_create(dev_priv, new_size,
-                           VMW_BO_DOMAIN_MOB, VMW_BO_DOMAIN_MOB,
-                           true, true, &buf);
+       ret = vmw_bo_create(dev_priv, &bo_params, &buf);
        if (ret) {
                DRM_ERROR("Failed initializing new cotable MOB.\n");
                goto out_done;
        }
 
-       bo = &buf->base;
+       bo = &buf->tbo;
        WARN_ON_ONCE(ttm_bo_reserve(bo, false, true, NULL));
 
        ret = ttm_bo_wait(old_bo, false, false);
@@ -477,8 +482,8 @@ static int vmw_cotable_resize(struct vmw_resource *res, size_t new_size)
        }
 
        vmw_resource_mob_detach(res);
-       res->backup = buf;
-       res->backup_size = new_size;
+       res->guest_memory_bo = buf;
+       res->guest_memory_size = new_size;
        vcotbl->size_read_back = cur_size_read_back;
 
        /*
@@ -488,8 +493,8 @@ static int vmw_cotable_resize(struct vmw_resource *res, size_t new_size)
        ret = vmw_cotable_unscrub(res);
        if (ret) {
                DRM_ERROR("Failed switching COTable backup buffer.\n");
-               res->backup = old_buf;
-               res->backup_size = old_size;
+               res->guest_memory_bo = old_buf;
+               res->guest_memory_size = old_size;
                vcotbl->size_read_back = old_size_read_back;
                vmw_resource_mob_attach(res);
                goto out_wait;
@@ -504,7 +509,7 @@ static int vmw_cotable_resize(struct vmw_resource *res, size_t new_size)
        if (unlikely(ret))
                goto out_wait;
 
-       /* Release the pin acquired in vmw_bo_init */
+       /* Release the pin acquired in vmw_bo_create */
        ttm_bo_unpin(bo);
 
        MKS_STAT_TIME_POP(MKSSTAT_KERN_COTABLE_RESIZE);
@@ -539,7 +544,7 @@ out_done:
 static int vmw_cotable_create(struct vmw_resource *res)
 {
        struct vmw_cotable *vcotbl = vmw_cotable(res);
-       size_t new_size = res->backup_size;
+       size_t new_size = res->guest_memory_size;
        size_t needed_size;
        int ret;
 
@@ -548,7 +553,7 @@ static int vmw_cotable_create(struct vmw_resource *res)
        while (needed_size > new_size)
                new_size *= 2;
 
-       if (likely(new_size <= res->backup_size)) {
+       if (likely(new_size <= res->guest_memory_size)) {
                if (vcotbl->scrubbed && vmw_resource_mob_attached(res)) {
                        ret = vmw_cotable_unscrub(res);
                        if (ret)
@@ -612,12 +617,12 @@ struct vmw_resource *vmw_cotable_alloc(struct vmw_private *dev_priv,
 
        INIT_LIST_HEAD(&vcotbl->resource_list);
        vcotbl->res.id = type;
-       vcotbl->res.backup_size = PAGE_SIZE;
+       vcotbl->res.guest_memory_size = PAGE_SIZE;
        num_entries = PAGE_SIZE / co_info[type].size;
        if (num_entries < co_info[type].min_initial_entries) {
-               vcotbl->res.backup_size = co_info[type].min_initial_entries *
+               vcotbl->res.guest_memory_size = co_info[type].min_initial_entries *
                        co_info[type].size;
-               vcotbl->res.backup_size = PFN_ALIGN(vcotbl->res.backup_size);
+               vcotbl->res.guest_memory_size = PFN_ALIGN(vcotbl->res.guest_memory_size);
        }
 
        vcotbl->scrubbed = true;
index d6300aa970f714d9ffa38c9a421d0041c459c21f..2588615a2a38ac666161c9a7acfcc4173f10685a 100644 (file)
@@ -391,23 +391,28 @@ static int vmw_dummy_query_bo_create(struct vmw_private *dev_priv)
        struct ttm_bo_kmap_obj map;
        volatile SVGA3dQueryResult *result;
        bool dummy;
+       struct vmw_bo_params bo_params = {
+               .domain = VMW_BO_DOMAIN_SYS,
+               .busy_domain = VMW_BO_DOMAIN_SYS,
+               .bo_type = ttm_bo_type_kernel,
+               .size = PAGE_SIZE,
+               .pin = true
+       };
 
        /*
         * Create the vbo as pinned, so that a tryreserve will
         * immediately succeed. This is because we're the only
         * user of the bo currently.
         */
-       ret = vmw_bo_create(dev_priv, PAGE_SIZE,
-                           VMW_BO_DOMAIN_SYS, VMW_BO_DOMAIN_SYS,
-                           false, true, &vbo);
+       ret = vmw_bo_create(dev_priv, &bo_params, &vbo);
        if (unlikely(ret != 0))
                return ret;
 
-       ret = ttm_bo_reserve(&vbo->base, false, true, NULL);
+       ret = ttm_bo_reserve(&vbo->tbo, false, true, NULL);
        BUG_ON(ret != 0);
        vmw_bo_pin_reserved(vbo, true);
 
-       ret = ttm_bo_kmap(&vbo->base, 0, 1, &map);
+       ret = ttm_bo_kmap(&vbo->tbo, 0, 1, &map);
        if (likely(ret == 0)) {
                result = ttm_kmap_obj_virtual(&map, &dummy);
                result->totalSize = sizeof(*result);
@@ -416,7 +421,7 @@ static int vmw_dummy_query_bo_create(struct vmw_private *dev_priv)
                ttm_bo_kunmap(&map);
        }
        vmw_bo_pin_reserved(vbo, false);
-       ttm_bo_unreserve(&vbo->base);
+       ttm_bo_unreserve(&vbo->tbo);
 
        if (unlikely(ret != 0)) {
                DRM_ERROR("Dummy query buffer map failed.\n");
index f903f78097a9f38c5b6fd7a1620f68ca77e7d1b9..fb8f0c0642c05eb2c21d1f83c13d37a673beebfa 100644 (file)
@@ -142,21 +142,23 @@ struct vmw_res_func;
  * @kref: For refcounting.
  * @dev_priv: Pointer to the device private for this resource. Immutable.
  * @id: Device id. Protected by @dev_priv::resource_lock.
- * @backup_size: Backup buffer size. Immutable.
- * @res_dirty: Resource contains data not yet in the backup buffer. Protected
- * by resource reserved.
- * @backup_dirty: Backup buffer contains data not yet in the HW resource.
+ * @guest_memory_size: Guest memory buffer size. Immutable.
+ * @res_dirty: Resource contains data not yet in the guest memory buffer.
  * Protected by resource reserved.
+ * @guest_memory_dirty: Guest memory buffer contains data not yet in the HW
+ * resource. Protected by resource reserved.
  * @coherent: Emulate coherency by tracking vm accesses.
- * @backup: The backup buffer if any. Protected by resource reserved.
- * @backup_offset: Offset into the backup buffer if any. Protected by resource
- * reserved. Note that only a few resource types can have a @backup_offset
- * different from zero.
+ * @guest_memory_bo: The guest memory buffer if any. Protected by resource
+ * reserved.
+ * @guest_memory_offset: Offset into the guest memory buffer if any. Protected
+ * by resource reserved. Note that only a few resource types can have a
+ * @guest_memory_offset different from zero.
  * @pin_count: The pin count for this resource. A pinned resource has a
  * pin-count greater than zero. It is not on the resource LRU lists and its
- * backup buffer is pinned. Hence it can't be evicted.
+ * guest memory buffer is pinned. Hence it can't be evicted.
  * @func: Method vtable for this resource. Immutable.
- * @mob_node; Node for the MOB backup rbtree. Protected by @backup reserved.
+ * @mob_node; Node for the MOB guest memory rbtree. Protected by
+ * @guest_memory_bo reserved.
  * @lru_head: List head for the LRU list. Protected by @dev_priv::resource_lock.
  * @binding_head: List head for the context binding list. Protected by
  * the @dev_priv::binding_mutex
@@ -165,18 +167,19 @@ struct vmw_res_func;
  * resource destruction.
  */
 struct vmw_bo;
+struct vmw_bo;
 struct vmw_resource_dirty;
 struct vmw_resource {
        struct kref kref;
        struct vmw_private *dev_priv;
        int id;
        u32 used_prio;
-       unsigned long backup_size;
+       unsigned long guest_memory_size;
        u32 res_dirty : 1;
-       u32 backup_dirty : 1;
+       u32 guest_memory_dirty : 1;
        u32 coherent : 1;
-       struct vmw_bo *backup;
-       unsigned long backup_offset;
+       struct vmw_bo *guest_memory_bo;
+       unsigned long guest_memory_offset;
        unsigned long pin_count;
        const struct vmw_res_func *func;
        struct rb_node mob_node;
@@ -467,7 +470,7 @@ struct vmw_otable_batch {
        unsigned num_otables;
        struct vmw_otable *otables;
        struct vmw_resource *context;
-       struct ttm_buffer_object *otable_bo;
+       struct vmw_bo *otable_bo;
 };
 
 enum {
@@ -662,6 +665,11 @@ static inline struct vmw_private *vmw_priv(struct drm_device *dev)
        return (struct vmw_private *)dev->dev_private;
 }
 
+static inline struct vmw_private *vmw_priv_from_ttm(struct ttm_device *bdev)
+{
+       return container_of(bdev, struct vmw_private, bdev);
+}
+
 static inline struct vmw_fpriv *vmw_fpriv(struct drm_file *file_priv)
 {
        return (struct vmw_fpriv *)file_priv->driver_priv;
@@ -814,9 +822,9 @@ extern int vmw_user_stream_lookup(struct vmw_private *dev_priv,
 extern void vmw_resource_unreserve(struct vmw_resource *res,
                                   bool dirty_set,
                                   bool dirty,
-                                  bool switch_backup,
-                                  struct vmw_bo *new_backup,
-                                  unsigned long new_backup_offset);
+                                  bool switch_guest_memory,
+                                  struct vmw_bo *new_guest_memory,
+                                  unsigned long new_guest_memory_offset);
 extern void vmw_query_move_notify(struct ttm_buffer_object *bo,
                                  struct ttm_resource *old_mem,
                                  struct ttm_resource *new_mem);
@@ -929,16 +937,15 @@ vmw_is_cursor_bypass3_enabled(const struct vmw_private *dev_priv)
 
 extern const size_t vmw_tt_size;
 extern struct ttm_placement vmw_vram_placement;
-extern struct ttm_placement vmw_vram_sys_placement;
 extern struct ttm_placement vmw_vram_gmr_placement;
 extern struct ttm_placement vmw_sys_placement;
-extern struct ttm_placement vmw_mob_placement;
 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,
-                                     unsigned long bo_size,
-                                     struct ttm_buffer_object **bo_p);
+int vmw_bo_create_and_populate(struct vmw_private *dev_priv,
+                              size_t bo_size,
+                              u32 domain,
+                              struct vmw_bo **bo_p);
 
 extern void vmw_piter_start(struct vmw_piter *viter,
                            const struct vmw_sg_table *vsgt,
index d135db90b4a5b6e0c5f49d63b0c734ffddcc9d7c..6d1b46c23719bc170d0a4cce083b8bfb8679179f 100644 (file)
@@ -722,7 +722,7 @@ static int vmw_rebind_all_dx_query(struct vmw_resource *ctx_res)
        cmd->header.id = SVGA_3D_CMD_DX_BIND_ALL_QUERY;
        cmd->header.size = sizeof(cmd->body);
        cmd->body.cid = ctx_res->id;
-       cmd->body.mobid = dx_query_mob->base.resource->start;
+       cmd->body.mobid = dx_query_mob->tbo.resource->start;
        vmw_cmd_commit(dev_priv, sizeof(*cmd));
 
        vmw_context_bind_dx_query(ctx_res, dx_query_mob);
@@ -1033,7 +1033,7 @@ static int vmw_query_bo_switch_prepare(struct vmw_private *dev_priv,
 
        if (unlikely(new_query_bo != sw_context->cur_query_bo)) {
 
-               if (unlikely(PFN_UP(new_query_bo->base.resource->size) > 4)) {
+               if (unlikely(PFN_UP(new_query_bo->tbo.resource->size) > 4)) {
                        VMW_DEBUG_USER("Query buffer too large.\n");
                        return -EINVAL;
                }
@@ -1164,7 +1164,7 @@ static int vmw_translate_mob_ptr(struct vmw_private *dev_priv,
        }
        vmw_bo_placement_set(vmw_bo, VMW_BO_DOMAIN_MOB, VMW_BO_DOMAIN_MOB);
        ret = vmw_validation_add_bo(sw_context->ctx, vmw_bo);
-       ttm_bo_put(&vmw_bo->base);
+       ttm_bo_put(&vmw_bo->tbo);
        if (unlikely(ret != 0))
                return ret;
 
@@ -1220,7 +1220,7 @@ static int vmw_translate_guest_ptr(struct vmw_private *dev_priv,
        vmw_bo_placement_set(vmw_bo, VMW_BO_DOMAIN_GMR | VMW_BO_DOMAIN_VRAM,
                             VMW_BO_DOMAIN_GMR | VMW_BO_DOMAIN_VRAM);
        ret = vmw_validation_add_bo(sw_context->ctx, vmw_bo);
-       ttm_bo_put(&vmw_bo->base);
+       ttm_bo_put(&vmw_bo->tbo);
        if (unlikely(ret != 0))
                return ret;
 
@@ -1533,7 +1533,7 @@ static int vmw_cmd_dma(struct vmw_private *dev_priv,
                return ret;
 
        /* Make sure DMA doesn't cross BO boundaries. */
-       bo_size = vmw_bo->base.base.size;
+       bo_size = vmw_bo->tbo.base.size;
        if (unlikely(cmd->body.guest.ptr.offset > bo_size)) {
                VMW_DEBUG_USER("Invalid DMA offset.\n");
                return -EINVAL;
@@ -1556,7 +1556,7 @@ static int vmw_cmd_dma(struct vmw_private *dev_priv,
 
        srf = vmw_res_to_srf(sw_context->res_cache[vmw_res_surface].res);
 
-       vmw_kms_cursor_snoop(srf, sw_context->fp->tfile, &vmw_bo->base, header);
+       vmw_kms_cursor_snoop(srf, sw_context->fp->tfile, &vmw_bo->tbo, header);
 
        return 0;
 }
@@ -3759,7 +3759,7 @@ static void vmw_apply_relocations(struct vmw_sw_context *sw_context)
        struct ttm_buffer_object *bo;
 
        list_for_each_entry(reloc, &sw_context->bo_relocations, head) {
-               bo = &reloc->vbo->base;
+               bo = &reloc->vbo->tbo;
                switch (bo->resource->mem_type) {
                case TTM_PL_VRAM:
                        reloc->location->offset += bo->resource->start << PAGE_SHIFT;
index 5f383578a3200224f1b5a0e34d3793c2df1ba3d3..f042e22b8b59b0cfab14765d8b00ef2a460c0eb8 100644 (file)
@@ -33,9 +33,8 @@
 static void vmw_gem_object_free(struct drm_gem_object *gobj)
 {
        struct ttm_buffer_object *bo = drm_gem_ttm_of_gem(gobj);
-       if (bo) {
+       if (bo)
                ttm_bo_put(bo);
-       }
 }
 
 static int vmw_gem_object_open(struct drm_gem_object *obj,
@@ -119,19 +118,23 @@ int vmw_gem_object_create_with_handle(struct vmw_private *dev_priv,
                                      struct vmw_bo **p_vbo)
 {
        int ret;
+       struct vmw_bo_params params = {
+               .domain = (dev_priv->has_mob) ? VMW_BO_DOMAIN_SYS : VMW_BO_DOMAIN_VRAM,
+               .busy_domain = VMW_BO_DOMAIN_SYS,
+               .bo_type = ttm_bo_type_device,
+               .size = size,
+               .pin = false
+       };
 
-       ret = vmw_bo_create(dev_priv, size,
-                           (dev_priv->has_mob) ? VMW_BO_DOMAIN_SYS : VMW_BO_DOMAIN_VRAM,
-                           VMW_BO_DOMAIN_SYS,
-                           true, false, p_vbo);
+       ret = vmw_bo_create(dev_priv, &params, p_vbo);
 
-       (*p_vbo)->base.base.funcs = &vmw_gem_object_funcs;
+       (*p_vbo)->tbo.base.funcs = &vmw_gem_object_funcs;
        if (ret != 0)
                goto out_no_bo;
 
-       ret = drm_gem_handle_create(filp, &(*p_vbo)->base.base, handle);
+       ret = drm_gem_handle_create(filp, &(*p_vbo)->tbo.base, handle);
        /* drop reference from allocate - handle holds it now */
-       drm_gem_object_put(&(*p_vbo)->base.base);
+       drm_gem_object_put(&(*p_vbo)->tbo.base);
 out_no_bo:
        return ret;
 }
@@ -155,7 +158,7 @@ int vmw_gem_object_create_ioctl(struct drm_device *dev, void *data,
                goto out_no_bo;
 
        rep->handle = handle;
-       rep->map_handle = drm_vma_node_offset_addr(&vbo->base.base.vma_node);
+       rep->map_handle = drm_vma_node_offset_addr(&vbo->tbo.base.vma_node);
        rep->cur_gmr_id = handle;
        rep->cur_gmr_offset = 0;
 out_no_bo:
@@ -169,7 +172,7 @@ static void vmw_bo_print_info(int id, struct vmw_bo *bo, struct seq_file *m)
        const char *placement;
        const char *type;
 
-       switch (bo->base.resource->mem_type) {
+       switch (bo->tbo.resource->mem_type) {
        case TTM_PL_SYSTEM:
                placement = " CPU";
                break;
@@ -190,7 +193,7 @@ static void vmw_bo_print_info(int id, struct vmw_bo *bo, struct seq_file *m)
                break;
        }
 
-       switch (bo->base.type) {
+       switch (bo->tbo.type) {
        case ttm_bo_type_device:
                type = "device";
                break;
@@ -206,12 +209,12 @@ static void vmw_bo_print_info(int id, struct vmw_bo *bo, struct seq_file *m)
        }
 
        seq_printf(m, "\t\t0x%08x: %12zu bytes %s, type = %s",
-                  id, bo->base.base.size, placement, type);
+                  id, bo->tbo.base.size, placement, type);
        seq_printf(m, ", priority = %u, pin_count = %u, GEM refs = %d, TTM refs = %d",
-                  bo->base.priority,
-                  bo->base.pin_count,
-                  kref_read(&bo->base.base.refcount),
-                  kref_read(&bo->base.kref));
+                  bo->tbo.priority,
+                  bo->tbo.pin_count,
+                  kref_read(&bo->tbo.base.refcount),
+                  kref_read(&bo->tbo.kref));
        seq_puts(m, "\n");
 }
 
index d975c0a818c79167ed47b77d231a97972234a99e..8659de9d23f354be7611cf480169bbabfae645f2 100644 (file)
@@ -153,9 +153,8 @@ static void vmw_cursor_update_mob(struct vmw_private *dev_priv,
        SVGAGBCursorHeader *header;
        SVGAGBAlphaCursorHeader *alpha_header;
        const u32 image_size = width * height * sizeof(*image);
-       bool is_iomem;
 
-       header = ttm_kmap_obj_virtual(&vps->cursor.map, &is_iomem);
+       header = vmw_bo_map_and_cache(vps->cursor.bo);
        alpha_header = &header->header.alphaHeader;
 
        memset(header, 0, sizeof(*header));
@@ -170,7 +169,7 @@ static void vmw_cursor_update_mob(struct vmw_private *dev_priv,
 
        memcpy(header + 1, image, image_size);
        vmw_write(dev_priv, SVGA_REG_CURSOR_MOBID,
-                 vps->cursor.bo->resource->start);
+                 vps->cursor.bo->tbo.resource->start);
 }
 
 
@@ -188,7 +187,7 @@ static u32 *vmw_du_cursor_plane_acquire_image(struct vmw_plane_state *vps)
        bool is_iomem;
        if (vps->surf) {
                if (vps->surf_mapped)
-                       return vmw_bo_map_and_cache(vps->surf->res.backup);
+                       return vmw_bo_map_and_cache(vps->surf->res.guest_memory_bo);
                return vps->surf->snooper.image;
        } else if (vps->bo)
                return ttm_kmap_obj_virtual(&vps->bo->map, &is_iomem);
@@ -223,15 +222,13 @@ static bool vmw_du_cursor_plane_has_changed(struct vmw_plane_state *old_vps,
        return changed;
 }
 
-static void vmw_du_destroy_cursor_mob(struct ttm_buffer_object **bo)
+static void vmw_du_destroy_cursor_mob(struct vmw_bo **vbo)
 {
-       if (!(*bo))
+       if (!(*vbo))
                return;
 
-       ttm_bo_unpin(*bo);
-       ttm_bo_put(*bo);
-       kfree(*bo);
-       *bo = NULL;
+       ttm_bo_unpin(&(*vbo)->tbo);
+       vmw_bo_unreference(vbo);
 }
 
 static void vmw_du_put_cursor_mob(struct vmw_cursor_plane *vcp,
@@ -255,8 +252,8 @@ static void vmw_du_put_cursor_mob(struct vmw_cursor_plane *vcp,
 
        /* Cache is full: See if this mob is bigger than an existing mob. */
        for (i = 0; i < ARRAY_SIZE(vcp->cursor_mobs); i++) {
-               if (vcp->cursor_mobs[i]->base.size <
-                   vps->cursor.bo->base.size) {
+               if (vcp->cursor_mobs[i]->tbo.base.size <
+                   vps->cursor.bo->tbo.base.size) {
                        vmw_du_destroy_cursor_mob(&vcp->cursor_mobs[i]);
                        vcp->cursor_mobs[i] = vps->cursor.bo;
                        vps->cursor.bo = NULL;
@@ -289,7 +286,7 @@ static int vmw_du_get_cursor_mob(struct vmw_cursor_plane *vcp,
                return -EINVAL;
 
        if (vps->cursor.bo) {
-               if (vps->cursor.bo->base.size >= size)
+               if (vps->cursor.bo->tbo.base.size >= size)
                        return 0;
                vmw_du_put_cursor_mob(vcp, vps);
        }
@@ -297,26 +294,27 @@ static int vmw_du_get_cursor_mob(struct vmw_cursor_plane *vcp,
        /* Look for an unused mob in the cache. */
        for (i = 0; i < ARRAY_SIZE(vcp->cursor_mobs); i++) {
                if (vcp->cursor_mobs[i] &&
-                   vcp->cursor_mobs[i]->base.size >= size) {
+                   vcp->cursor_mobs[i]->tbo.base.size >= size) {
                        vps->cursor.bo = vcp->cursor_mobs[i];
                        vcp->cursor_mobs[i] = NULL;
                        return 0;
                }
        }
        /* Create a new mob if we can't find an existing one. */
-       ret = vmw_bo_create_kernel(dev_priv, size, &vmw_mob_placement,
-                                  &vps->cursor.bo);
+       ret = vmw_bo_create_and_populate(dev_priv, size,
+                                        VMW_BO_DOMAIN_MOB,
+                                        &vps->cursor.bo);
 
        if (ret != 0)
                return ret;
 
        /* Fence the mob creation so we are guarateed to have the mob */
-       ret = ttm_bo_reserve(vps->cursor.bo, false, false, NULL);
+       ret = ttm_bo_reserve(&vps->cursor.bo->tbo, false, false, NULL);
        if (ret != 0)
                goto teardown;
 
-       vmw_bo_fence_single(vps->cursor.bo, NULL);
-       ttm_bo_unreserve(vps->cursor.bo);
+       vmw_bo_fence_single(&vps->cursor.bo->tbo, NULL);
+       ttm_bo_unreserve(&vps->cursor.bo->tbo);
        return 0;
 
 teardown:
@@ -574,39 +572,30 @@ vmw_du_cursor_plane_map_cm(struct vmw_plane_state *vps)
 {
        int ret;
        u32 size = vmw_du_cursor_mob_size(vps->base.crtc_w, vps->base.crtc_h);
-       struct ttm_buffer_object *bo = vps->cursor.bo;
+       struct ttm_buffer_object *bo;
 
-       if (!bo)
+       if (!vps->cursor.bo)
                return -EINVAL;
 
+       bo = &vps->cursor.bo->tbo;
+
        if (bo->base.size < size)
                return -EINVAL;
 
-       if (vps->cursor.mapped)
+       if (vps->cursor.bo->map.virtual)
                return 0;
 
        ret = ttm_bo_reserve(bo, false, false, NULL);
-
        if (unlikely(ret != 0))
                return -ENOMEM;
 
-       ret = ttm_bo_kmap(bo, 0, PFN_UP(size), &vps->cursor.map);
-
-       /*
-        * We just want to try to get mob bind to finish
-        * so that the first write to SVGA_REG_CURSOR_MOBID
-        * is done with a buffer that the device has already
-        * seen
-        */
-       (void) ttm_bo_wait(bo, false, false);
+       vmw_bo_map_and_cache(vps->cursor.bo);
 
        ttm_bo_unreserve(bo);
 
        if (unlikely(ret != 0))
                return -ENOMEM;
 
-       vps->cursor.mapped = true;
-
        return 0;
 }
 
@@ -623,19 +612,15 @@ static int
 vmw_du_cursor_plane_unmap_cm(struct vmw_plane_state *vps)
 {
        int ret = 0;
-       struct ttm_buffer_object *bo = vps->cursor.bo;
+       struct vmw_bo *vbo = vps->cursor.bo;
 
-       if (!vps->cursor.mapped)
+       if (!vbo || !vbo->map.virtual)
                return 0;
 
-       if (!bo)
-               return 0;
-
-       ret = ttm_bo_reserve(bo, true, false, NULL);
+       ret = ttm_bo_reserve(&vbo->tbo, true, false, NULL);
        if (likely(ret == 0)) {
-               ttm_bo_kunmap(&vps->cursor.map);
-               ttm_bo_unreserve(bo);
-               vps->cursor.mapped = false;
+               vmw_bo_unmap(vbo);
+               ttm_bo_unreserve(&vbo->tbo);
        }
 
        return ret;
@@ -661,16 +646,16 @@ vmw_du_cursor_plane_cleanup_fb(struct drm_plane *plane,
        bool is_iomem;
 
        if (vps->surf_mapped) {
-               vmw_bo_unmap(vps->surf->res.backup);
+               vmw_bo_unmap(vps->surf->res.guest_memory_bo);
                vps->surf_mapped = false;
        }
 
        if (vps->bo && ttm_kmap_obj_virtual(&vps->bo->map, &is_iomem)) {
-               const int ret = ttm_bo_reserve(&vps->bo->base, true, false, NULL);
+               const int ret = ttm_bo_reserve(&vps->bo->tbo, true, false, NULL);
 
                if (likely(ret == 0)) {
                        ttm_bo_kunmap(&vps->bo->map);
-                       ttm_bo_unreserve(&vps->bo->base);
+                       ttm_bo_unreserve(&vps->bo->tbo);
                }
        }
 
@@ -736,26 +721,26 @@ vmw_du_cursor_plane_prepare_fb(struct drm_plane *plane,
                 * reserve the ttm_buffer_object first which
                 * vmw_bo_map_and_cache() omits.
                 */
-               ret = ttm_bo_reserve(&vps->bo->base, true, false, NULL);
+               ret = ttm_bo_reserve(&vps->bo->tbo, true, false, NULL);
 
                if (unlikely(ret != 0))
                        return -ENOMEM;
 
-               ret = ttm_bo_kmap(&vps->bo->base, 0, PFN_UP(size), &vps->bo->map);
+               ret = ttm_bo_kmap(&vps->bo->tbo, 0, PFN_UP(size), &vps->bo->map);
 
-               ttm_bo_unreserve(&vps->bo->base);
+               ttm_bo_unreserve(&vps->bo->tbo);
 
                if (unlikely(ret != 0))
                        return -ENOMEM;
-       } else if (vps->surf && !vps->bo && vps->surf->res.backup) {
+       } else if (vps->surf && !vps->bo && vps->surf->res.guest_memory_bo) {
 
                WARN_ON(vps->surf->snooper.image);
-               ret = ttm_bo_reserve(&vps->surf->res.backup->base, true, false,
+               ret = ttm_bo_reserve(&vps->surf->res.guest_memory_bo->tbo, true, false,
                                     NULL);
                if (unlikely(ret != 0))
                        return -ENOMEM;
-               vmw_bo_map_and_cache(vps->surf->res.backup);
-               ttm_bo_unreserve(&vps->surf->res.backup->base);
+               vmw_bo_map_and_cache(vps->surf->res.guest_memory_bo);
+               ttm_bo_unreserve(&vps->surf->res.guest_memory_bo->tbo);
                vps->surf_mapped = true;
        }
 
@@ -926,7 +911,7 @@ int vmw_du_cursor_plane_atomic_check(struct drm_plane *plane,
                WARN_ON(!surface);
 
                if (!surface ||
-                   (!surface->snooper.image && !surface->res.backup)) {
+                   (!surface->snooper.image && !surface->res.guest_memory_bo)) {
                        DRM_ERROR("surface not suitable for cursor\n");
                        return -EINVAL;
                }
@@ -1397,7 +1382,7 @@ static int vmw_framebuffer_bo_create_handle(struct drm_framebuffer *fb,
        struct vmw_framebuffer_bo *vfbd =
                        vmw_framebuffer_to_vfbd(fb);
 
-       return drm_gem_handle_create(file_priv, &vfbd->buffer->base.base, handle);
+       return drm_gem_handle_create(file_priv, &vfbd->buffer->tbo.base, handle);
 }
 
 static void vmw_framebuffer_bo_destroy(struct drm_framebuffer *framebuffer)
@@ -1546,9 +1531,9 @@ static int vmw_create_bo_proxy(struct drm_device *dev,
        /* Reserve and switch the backing mob. */
        mutex_lock(&res->dev_priv->cmdbuf_mutex);
        (void) vmw_resource_reserve(res, false, true);
-       vmw_bo_unreference(&res->backup);
-       res->backup = vmw_bo_reference(bo_mob);
-       res->backup_offset = 0;
+       vmw_bo_unreference(&res->guest_memory_bo);
+       res->guest_memory_bo = vmw_bo_reference(bo_mob);
+       res->guest_memory_offset = 0;
        vmw_resource_unreserve(res, false, false, false, NULL, 0);
        mutex_unlock(&res->dev_priv->cmdbuf_mutex);
 
@@ -1570,7 +1555,7 @@ static int vmw_kms_new_framebuffer_bo(struct vmw_private *dev_priv,
        int ret;
 
        requested_size = mode_cmd->height * mode_cmd->pitches[0];
-       if (unlikely(requested_size > bo->base.base.size)) {
+       if (unlikely(requested_size > bo->tbo.base.size)) {
                DRM_ERROR("Screen buffer object size is too small "
                          "for requested mode.\n");
                return -EINVAL;
@@ -1591,7 +1576,7 @@ static int vmw_kms_new_framebuffer_bo(struct vmw_private *dev_priv,
                goto out_err1;
        }
 
-       vfbd->base.base.obj[0] = &bo->base.base;
+       vfbd->base.base.obj[0] = &bo->tbo.base;
        drm_helper_mode_fill_fb_struct(dev, &vfbd->base.base, mode_cmd);
        vfbd->base.bo = true;
        vfbd->buffer = vmw_bo_reference(bo);
index 1bdf601e7c3528432002896683a37850a486857a..3de7b4b6a23097a65831c3bc1bd3fd6cc972a546 100644 (file)
@@ -270,9 +270,7 @@ struct vmw_crtc_state {
 };
 
 struct vmw_cursor_plane_state {
-       struct ttm_buffer_object *bo;
-       struct ttm_bo_kmap_obj map;
-       bool mapped;
+       struct vmw_bo *bo;
        s32 hotspot_x;
        s32 hotspot_y;
 };
@@ -343,7 +341,7 @@ struct vmw_connector_state {
 struct vmw_cursor_plane {
        struct drm_plane base;
 
-       struct ttm_buffer_object *cursor_mobs[3];
+       struct vmw_bo *cursor_mobs[3];
 };
 
 /**
index b77fe0bc18a76e9035ea1575e79992cc75667641..c0e42f2ed144c2e87f7016e3c5618a7f9dd3e8e4 100644 (file)
@@ -147,7 +147,7 @@ static int vmw_ldu_fb_pin(struct vmw_framebuffer *vfb)
        int ret;
 
        buf = vfb->bo ?  vmw_framebuffer_to_vfbd(&vfb->base)->buffer :
-               vmw_framebuffer_to_vfbs(&vfb->base)->surface->res.backup;
+               vmw_framebuffer_to_vfbs(&vfb->base)->surface->res.guest_memory_bo;
 
        if (!buf)
                return 0;
@@ -169,7 +169,7 @@ static int vmw_ldu_fb_unpin(struct vmw_framebuffer *vfb)
        struct vmw_bo *buf;
 
        buf = vfb->bo ?  vmw_framebuffer_to_vfbd(&vfb->base)->buffer :
-               vmw_framebuffer_to_vfbs(&vfb->base)->surface->res.backup;
+               vmw_framebuffer_to_vfbs(&vfb->base)->surface->res.guest_memory_bo;
 
        if (WARN_ON(!buf))
                return 0;
index 6b3f53b533dc26fcc2db086e8a24c36e3dc140d1..7055cbefc768407b0f910af1fb2a8e5981e8babd 100644 (file)
@@ -51,7 +51,7 @@
  * @pt_root_page    DMA address of the level 0 page of the page table.
  */
 struct vmw_mob {
-       struct ttm_buffer_object *pt_bo;
+       struct vmw_bo *pt_bo;
        unsigned long num_pages;
        unsigned pt_level;
        dma_addr_t pt_root_page;
@@ -204,7 +204,7 @@ static void vmw_takedown_otable_base(struct vmw_private *dev_priv,
        if (otable->page_table == NULL)
                return;
 
-       bo = otable->page_table->pt_bo;
+       bo = &otable->page_table->pt_bo->tbo;
        cmd = VMW_CMD_RESERVE(dev_priv, sizeof(*cmd));
        if (unlikely(cmd == NULL))
                return;
@@ -252,7 +252,9 @@ static int vmw_otable_batch_setup(struct vmw_private *dev_priv,
                bo_size += otables[i].size;
        }
 
-       ret = vmw_bo_create_and_populate(dev_priv, bo_size, &batch->otable_bo);
+       ret = vmw_bo_create_and_populate(dev_priv, bo_size,
+                                        VMW_BO_DOMAIN_WAITABLE_SYS,
+                                        &batch->otable_bo);
        if (unlikely(ret != 0))
                return ret;
 
@@ -261,7 +263,8 @@ static int vmw_otable_batch_setup(struct vmw_private *dev_priv,
                if (!batch->otables[i].enabled)
                        continue;
 
-               ret = vmw_setup_otable_base(dev_priv, i, batch->otable_bo,
+               ret = vmw_setup_otable_base(dev_priv, i,
+                                           &batch->otable_bo->tbo,
                                            offset,
                                            &otables[i]);
                if (unlikely(ret != 0))
@@ -278,8 +281,8 @@ out_no_setup:
                                                 &batch->otables[i]);
        }
 
-       vmw_bo_unpin_unlocked(batch->otable_bo);
-       ttm_bo_put(batch->otable_bo);
+       vmw_bo_unpin_unlocked(&batch->otable_bo->tbo);
+       ttm_bo_put(&batch->otable_bo->tbo);
        batch->otable_bo = NULL;
        return ret;
 }
@@ -330,7 +333,7 @@ static void vmw_otable_batch_takedown(struct vmw_private *dev_priv,
                               struct vmw_otable_batch *batch)
 {
        SVGAOTableType i;
-       struct ttm_buffer_object *bo = batch->otable_bo;
+       struct ttm_buffer_object *bo = &batch->otable_bo->tbo;
        int ret;
 
        for (i = 0; i < batch->num_otables; ++i)
@@ -345,8 +348,7 @@ static void vmw_otable_batch_takedown(struct vmw_private *dev_priv,
        ttm_bo_unpin(bo);
        ttm_bo_unreserve(bo);
 
-       ttm_bo_put(batch->otable_bo);
-       batch->otable_bo = NULL;
+       vmw_bo_unreference(&batch->otable_bo);
 }
 
 /*
@@ -414,7 +416,9 @@ static int vmw_mob_pt_populate(struct vmw_private *dev_priv,
 {
        BUG_ON(mob->pt_bo != NULL);
 
-       return vmw_bo_create_and_populate(dev_priv, mob->num_pages * PAGE_SIZE, &mob->pt_bo);
+       return vmw_bo_create_and_populate(dev_priv, mob->num_pages * PAGE_SIZE,
+                                         VMW_BO_DOMAIN_WAITABLE_SYS,
+                                         &mob->pt_bo);
 }
 
 /**
@@ -495,7 +499,7 @@ static void vmw_mob_pt_setup(struct vmw_mob *mob,
                             unsigned long num_data_pages)
 {
        unsigned long num_pt_pages = 0;
-       struct ttm_buffer_object *bo = mob->pt_bo;
+       struct ttm_buffer_object *bo = &mob->pt_bo->tbo;
        struct vmw_piter save_pt_iter = {0};
        struct vmw_piter pt_iter;
        const struct vmw_sg_table *vsgt;
@@ -532,9 +536,8 @@ static void vmw_mob_pt_setup(struct vmw_mob *mob,
 void vmw_mob_destroy(struct vmw_mob *mob)
 {
        if (mob->pt_bo) {
-               vmw_bo_unpin_unlocked(mob->pt_bo);
-               ttm_bo_put(mob->pt_bo);
-               mob->pt_bo = NULL;
+               vmw_bo_unpin_unlocked(&mob->pt_bo->tbo);
+               vmw_bo_unreference(&mob->pt_bo);
        }
        kfree(mob);
 }
@@ -553,7 +556,7 @@ void vmw_mob_unbind(struct vmw_private *dev_priv,
                SVGA3dCmdDestroyGBMob body;
        } *cmd;
        int ret;
-       struct ttm_buffer_object *bo = mob->pt_bo;
+       struct ttm_buffer_object *bo = &mob->pt_bo->tbo;
 
        if (bo) {
                ret = ttm_bo_reserve(bo, false, true, NULL);
@@ -645,9 +648,8 @@ int vmw_mob_bind(struct vmw_private *dev_priv,
 out_no_cmd_space:
        vmw_fifo_resource_dec(dev_priv);
        if (pt_set_up) {
-               vmw_bo_unpin_unlocked(mob->pt_bo);
-               ttm_bo_put(mob->pt_bo);
-               mob->pt_bo = NULL;
+               vmw_bo_unpin_unlocked(&mob->pt_bo->tbo);
+               vmw_bo_unreference(&mob->pt_bo);
        }
 
        return -ENOMEM;
index b415e86b26db94baf5bfe5d1565d1910ff5b19df..7bcda29a2897ddc73f84538af6a71840783543ed 100644 (file)
@@ -140,7 +140,7 @@ static int vmw_overlay_send_put(struct vmw_private *dev_priv,
        for (i = 0; i < num_items; i++)
                items[i].registerId = i;
 
-       vmw_bo_get_guest_ptr(&buf->base, &ptr);
+       vmw_bo_get_guest_ptr(&buf->tbo, &ptr);
        ptr.offset += arg->offset;
 
        items[SVGA_VIDEO_ENABLED].value     = true;
index c92ca6dabe3cc45249fe5c9321cda0324de3a900..74ff2812d66a1b9d08ae92db9f74bc58007b241e 100644 (file)
@@ -82,8 +82,8 @@ struct vmw_bo_dirty {
 static void vmw_bo_dirty_scan_pagetable(struct vmw_bo *vbo)
 {
        struct vmw_bo_dirty *dirty = vbo->dirty;
-       pgoff_t offset = drm_vma_node_start(&vbo->base.base.vma_node);
-       struct address_space *mapping = vbo->base.bdev->dev_mapping;
+       pgoff_t offset = drm_vma_node_start(&vbo->tbo.base.vma_node);
+       struct address_space *mapping = vbo->tbo.bdev->dev_mapping;
        pgoff_t num_marked;
 
        num_marked = clean_record_shared_mapping_range
@@ -120,23 +120,22 @@ static void vmw_bo_dirty_scan_pagetable(struct vmw_bo *vbo)
 static void vmw_bo_dirty_scan_mkwrite(struct vmw_bo *vbo)
 {
        struct vmw_bo_dirty *dirty = vbo->dirty;
-       unsigned long offset = drm_vma_node_start(&vbo->base.base.vma_node);
-       struct address_space *mapping = vbo->base.bdev->dev_mapping;
+       unsigned long offset = drm_vma_node_start(&vbo->tbo.base.vma_node);
+       struct address_space *mapping = vbo->tbo.bdev->dev_mapping;
        pgoff_t num_marked;
 
        if (dirty->end <= dirty->start)
                return;
 
-       num_marked = wp_shared_mapping_range(vbo->base.bdev->dev_mapping,
-                                       dirty->start + offset,
-                                       dirty->end - dirty->start);
+       num_marked = wp_shared_mapping_range(vbo->tbo.bdev->dev_mapping,
+                                            dirty->start + offset,
+                                            dirty->end - dirty->start);
 
        if (100UL * num_marked / dirty->bitmap_size >
-           VMW_DIRTY_PERCENTAGE) {
+           VMW_DIRTY_PERCENTAGE)
                dirty->change_count++;
-       } else {
+       else
                dirty->change_count = 0;
-       }
 
        if (dirty->change_count > VMW_DIRTY_NUM_CHANGE_TRIGGERS) {
                pgoff_t start = 0;
@@ -186,8 +185,8 @@ static void vmw_bo_dirty_pre_unmap(struct vmw_bo *vbo,
                                   pgoff_t start, pgoff_t end)
 {
        struct vmw_bo_dirty *dirty = vbo->dirty;
-       unsigned long offset = drm_vma_node_start(&vbo->base.base.vma_node);
-       struct address_space *mapping = vbo->base.bdev->dev_mapping;
+       unsigned long offset = drm_vma_node_start(&vbo->tbo.base.vma_node);
+       struct address_space *mapping = vbo->tbo.bdev->dev_mapping;
 
        if (dirty->method != VMW_BO_DIRTY_PAGETABLE || start >= end)
                return;
@@ -210,8 +209,8 @@ static void vmw_bo_dirty_pre_unmap(struct vmw_bo *vbo,
 void vmw_bo_dirty_unmap(struct vmw_bo *vbo,
                        pgoff_t start, pgoff_t end)
 {
-       unsigned long offset = drm_vma_node_start(&vbo->base.base.vma_node);
-       struct address_space *mapping = vbo->base.bdev->dev_mapping;
+       unsigned long offset = drm_vma_node_start(&vbo->tbo.base.vma_node);
+       struct address_space *mapping = vbo->tbo.bdev->dev_mapping;
 
        vmw_bo_dirty_pre_unmap(vbo, start, end);
        unmap_shared_mapping_range(mapping, (offset + start) << PAGE_SHIFT,
@@ -231,7 +230,7 @@ void vmw_bo_dirty_unmap(struct vmw_bo *vbo,
 int vmw_bo_dirty_add(struct vmw_bo *vbo)
 {
        struct vmw_bo_dirty *dirty = vbo->dirty;
-       pgoff_t num_pages = PFN_UP(vbo->base.resource->size);
+       pgoff_t num_pages = PFN_UP(vbo->tbo.resource->size);
        size_t size;
        int ret;
 
@@ -254,8 +253,8 @@ int vmw_bo_dirty_add(struct vmw_bo *vbo)
        if (num_pages < PAGE_SIZE / sizeof(pte_t)) {
                dirty->method = VMW_BO_DIRTY_PAGETABLE;
        } else {
-               struct address_space *mapping = vbo->base.bdev->dev_mapping;
-               pgoff_t offset = drm_vma_node_start(&vbo->base.base.vma_node);
+               struct address_space *mapping = vbo->tbo.bdev->dev_mapping;
+               pgoff_t offset = drm_vma_node_start(&vbo->tbo.base.vma_node);
 
                dirty->method = VMW_BO_DIRTY_MKWRITE;
 
@@ -307,11 +306,11 @@ void vmw_bo_dirty_release(struct vmw_bo *vbo)
  */
 void vmw_bo_dirty_transfer_to_res(struct vmw_resource *res)
 {
-       struct vmw_bo *vbo = res->backup;
+       struct vmw_bo *vbo = res->guest_memory_bo;
        struct vmw_bo_dirty *dirty = vbo->dirty;
        pgoff_t start, cur, end;
-       unsigned long res_start = res->backup_offset;
-       unsigned long res_end = res->backup_offset + res->backup_size;
+       unsigned long res_start = res->guest_memory_offset;
+       unsigned long res_end = res->guest_memory_offset + res->guest_memory_size;
 
        WARN_ON_ONCE(res_start & ~PAGE_MASK);
        res_start >>= PAGE_SHIFT;
@@ -352,9 +351,9 @@ void vmw_bo_dirty_transfer_to_res(struct vmw_resource *res)
  */
 void vmw_bo_dirty_clear_res(struct vmw_resource *res)
 {
-       unsigned long res_start = res->backup_offset;
-       unsigned long res_end = res->backup_offset + res->backup_size;
-       struct vmw_bo *vbo = res->backup;
+       unsigned long res_start = res->guest_memory_offset;
+       unsigned long res_end = res->guest_memory_offset + res->guest_memory_size;
+       struct vmw_bo *vbo = res->guest_memory_bo;
        struct vmw_bo_dirty *dirty = vbo->dirty;
 
        res_start >>= PAGE_SHIFT;
@@ -381,8 +380,7 @@ vm_fault_t vmw_bo_vm_mkwrite(struct vm_fault *vmf)
        vm_fault_t ret;
        unsigned long page_offset;
        unsigned int save_flags;
-       struct vmw_bo *vbo =
-               container_of(bo, typeof(*vbo), base);
+       struct vmw_bo *vbo = to_vmw_bo(&bo->base);
 
        /*
         * mkwrite() doesn't handle the VM_FAULT_RETRY return value correctly.
@@ -420,8 +418,7 @@ vm_fault_t vmw_bo_vm_fault(struct vm_fault *vmf)
        struct vm_area_struct *vma = vmf->vma;
        struct ttm_buffer_object *bo = (struct ttm_buffer_object *)
            vma->vm_private_data;
-       struct vmw_bo *vbo =
-               container_of(bo, struct vmw_bo, base);
+       struct vmw_bo *vbo = to_vmw_bo(&bo->base);
        pgoff_t num_prefault;
        pgprot_t prot;
        vm_fault_t ret;
index 6780a36e61710f302920b515d0a8be46f933ad9d..54e942df3b8e4425f4bce765adf3751a6d2c55fb 100644 (file)
  */
 void vmw_resource_mob_attach(struct vmw_resource *res)
 {
-       struct vmw_bo *backup = res->backup;
-       struct rb_node **new = &backup->res_tree.rb_node, *parent = NULL;
+       struct vmw_bo *gbo = res->guest_memory_bo;
+       struct rb_node **new = &gbo->res_tree.rb_node, *parent = NULL;
 
-       dma_resv_assert_held(res->backup->base.base.resv);
+       dma_resv_assert_held(gbo->tbo.base.resv);
        res->used_prio = (res->res_dirty) ? res->func->dirty_prio :
                res->func->prio;
 
@@ -52,14 +52,14 @@ void vmw_resource_mob_attach(struct vmw_resource *res)
                        container_of(*new, struct vmw_resource, mob_node);
 
                parent = *new;
-               new = (res->backup_offset < this->backup_offset) ?
+               new = (res->guest_memory_offset < this->guest_memory_offset) ?
                        &((*new)->rb_left) : &((*new)->rb_right);
        }
 
        rb_link_node(&res->mob_node, parent, new);
-       rb_insert_color(&res->mob_node, &backup->res_tree);
+       rb_insert_color(&res->mob_node, &gbo->res_tree);
 
-       vmw_bo_prio_add(backup, res->used_prio);
+       vmw_bo_prio_add(gbo, res->used_prio);
 }
 
 /**
@@ -68,13 +68,13 @@ void vmw_resource_mob_attach(struct vmw_resource *res)
  */
 void vmw_resource_mob_detach(struct vmw_resource *res)
 {
-       struct vmw_bo *backup = res->backup;
+       struct vmw_bo *gbo = res->guest_memory_bo;
 
-       dma_resv_assert_held(backup->base.base.resv);
+       dma_resv_assert_held(gbo->tbo.base.resv);
        if (vmw_resource_mob_attached(res)) {
-               rb_erase(&res->mob_node, &backup->res_tree);
+               rb_erase(&res->mob_node, &gbo->res_tree);
                RB_CLEAR_NODE(&res->mob_node);
-               vmw_bo_prio_del(backup, res->used_prio);
+               vmw_bo_prio_del(gbo, res->used_prio);
        }
 }
 
@@ -121,8 +121,8 @@ static void vmw_resource_release(struct kref *kref)
        spin_lock(&dev_priv->resource_lock);
        list_del_init(&res->lru_head);
        spin_unlock(&dev_priv->resource_lock);
-       if (res->backup) {
-               struct ttm_buffer_object *bo = &res->backup->base;
+       if (res->guest_memory_bo) {
+               struct ttm_buffer_object *bo = &res->guest_memory_bo->tbo;
 
                ret = ttm_bo_reserve(bo, false, false, NULL);
                BUG_ON(ret);
@@ -134,14 +134,14 @@ static void vmw_resource_release(struct kref *kref)
                        val_buf.num_shared = 0;
                        res->func->unbind(res, false, &val_buf);
                }
-               res->backup_dirty = false;
+               res->guest_memory_size = false;
                vmw_resource_mob_detach(res);
                if (res->dirty)
                        res->func->dirty_free(res);
                if (res->coherent)
-                       vmw_bo_dirty_release(res->backup);
+                       vmw_bo_dirty_release(res->guest_memory_bo);
                ttm_bo_unreserve(bo);
-               vmw_bo_unreference(&res->backup);
+               vmw_bo_unreference(&res->guest_memory_bo);
        }
 
        if (likely(res->hw_destroy != NULL)) {
@@ -224,9 +224,9 @@ int vmw_resource_init(struct vmw_private *dev_priv, struct vmw_resource *res,
        INIT_LIST_HEAD(&res->lru_head);
        INIT_LIST_HEAD(&res->binding_head);
        res->id = -1;
-       res->backup = NULL;
-       res->backup_offset = 0;
-       res->backup_dirty = false;
+       res->guest_memory_bo = NULL;
+       res->guest_memory_offset = 0;
+       res->guest_memory_dirty = false;
        res->res_dirty = false;
        res->coherent = false;
        res->used_prio = 3;
@@ -264,7 +264,7 @@ int vmw_user_resource_lookup_handle(struct vmw_private *dev_priv,
        int ret = -EINVAL;
 
        base = ttm_base_object_lookup(tfile, handle);
-       if (unlikely(base == NULL))
+       if (unlikely(!base))
                return -EINVAL;
 
        if (unlikely(ttm_base_object_type(base) != converter->object_type))
@@ -313,31 +313,36 @@ int vmw_user_lookup_handle(struct vmw_private *dev_priv,
 }
 
 /**
- * vmw_resource_buf_alloc - Allocate a backup buffer for a resource.
+ * vmw_resource_buf_alloc - Allocate a guest memory buffer for a resource.
  *
- * @res:            The resource for which to allocate a backup buffer.
+ * @res:            The resource for which to allocate a gbo buffer.
  * @interruptible:  Whether any sleeps during allocation should be
  *                  performed while interruptible.
  */
 static int vmw_resource_buf_alloc(struct vmw_resource *res,
                                  bool interruptible)
 {
-       unsigned long size = PFN_ALIGN(res->backup_size);
-       struct vmw_bo *backup;
+       unsigned long size = PFN_ALIGN(res->guest_memory_size);
+       struct vmw_bo *gbo;
+       struct vmw_bo_params bo_params = {
+               .domain = res->func->domain,
+               .busy_domain = res->func->busy_domain,
+               .bo_type = ttm_bo_type_device,
+               .size = res->guest_memory_size,
+               .pin = false
+       };
        int ret;
 
-       if (likely(res->backup)) {
-               BUG_ON(res->backup->base.base.size < size);
+       if (likely(res->guest_memory_bo)) {
+               BUG_ON(res->guest_memory_bo->tbo.base.size < size);
                return 0;
        }
 
-       ret = vmw_bo_create(res->dev_priv, res->backup_size,
-                           res->func->domain, res->func->busy_domain,
-                           interruptible, false, &backup);
+       ret = vmw_bo_create(res->dev_priv, &bo_params, &gbo);
        if (unlikely(ret != 0))
                goto out_no_bo;
 
-       res->backup = backup;
+       res->guest_memory_bo = gbo;
 
 out_no_bo:
        return ret;
@@ -369,13 +374,13 @@ static int vmw_resource_do_validate(struct vmw_resource *res,
        }
 
        if (func->bind &&
-           ((func->needs_backup && !vmw_resource_mob_attached(res) &&
-             val_buf->bo != NULL) ||
-            (!func->needs_backup && val_buf->bo != NULL))) {
+           ((func->needs_guest_memory && !vmw_resource_mob_attached(res) &&
+             val_buf->bo) ||
+            (!func->needs_guest_memory && val_buf->bo))) {
                ret = func->bind(res, val_buf);
                if (unlikely(ret != 0))
                        goto out_bind_failed;
-               if (func->needs_backup)
+               if (func->needs_guest_memory)
                        vmw_resource_mob_attach(res);
        }
 
@@ -385,11 +390,11 @@ static int vmw_resource_do_validate(struct vmw_resource *res,
         */
        if (func->dirty_alloc && vmw_resource_mob_attached(res) &&
            !res->coherent) {
-               if (res->backup->dirty && !res->dirty) {
+               if (res->guest_memory_bo->dirty && !res->dirty) {
                        ret = func->dirty_alloc(res);
                        if (ret)
                                return ret;
-               } else if (!res->backup->dirty && res->dirty) {
+               } else if (!res->guest_memory_bo->dirty && res->dirty) {
                        func->dirty_free(res);
                }
        }
@@ -400,12 +405,12 @@ static int vmw_resource_do_validate(struct vmw_resource *res,
         */
        if (res->dirty) {
                if (dirtying && !res->res_dirty) {
-                       pgoff_t start = res->backup_offset >> PAGE_SHIFT;
+                       pgoff_t start = res->guest_memory_offset >> PAGE_SHIFT;
                        pgoff_t end = __KERNEL_DIV_ROUND_UP
-                               (res->backup_offset + res->backup_size,
+                               (res->guest_memory_offset + res->guest_memory_size,
                                 PAGE_SIZE);
 
-                       vmw_bo_dirty_unmap(res->backup, start, end);
+                       vmw_bo_dirty_unmap(res->guest_memory_bo, start, end);
                }
 
                vmw_bo_dirty_transfer_to_res(res);
@@ -427,10 +432,10 @@ out_bind_failed:
  * @res:               Pointer to the struct vmw_resource to unreserve.
  * @dirty_set:         Change dirty status of the resource.
  * @dirty:             When changing dirty status indicates the new status.
- * @switch_backup:     Backup buffer has been switched.
- * @new_backup:        Pointer to new backup buffer if command submission
+ * @switch_guest_memory: Guest memory buffer has been switched.
+ * @new_guest_memory_bo: Pointer to new guest memory buffer if command submission
  *                     switched. May be NULL.
- * @new_backup_offset: New backup offset if @switch_backup is true.
+ * @new_guest_memory_offset: New gbo offset if @switch_guest_memory is true.
  *
  * Currently unreserving a resource means putting it back on the device's
  * resource lru list, so that it can be evicted if necessary.
@@ -438,42 +443,42 @@ out_bind_failed:
 void vmw_resource_unreserve(struct vmw_resource *res,
                            bool dirty_set,
                            bool dirty,
-                           bool switch_backup,
-                           struct vmw_bo *new_backup,
-                           unsigned long new_backup_offset)
+                           bool switch_guest_memory,
+                           struct vmw_bo *new_guest_memory_bo,
+                           unsigned long new_guest_memory_offset)
 {
        struct vmw_private *dev_priv = res->dev_priv;
 
        if (!list_empty(&res->lru_head))
                return;
 
-       if (switch_backup && new_backup != res->backup) {
-               if (res->backup) {
+       if (switch_guest_memory && new_guest_memory_bo != res->guest_memory_bo) {
+               if (res->guest_memory_bo) {
                        vmw_resource_mob_detach(res);
                        if (res->coherent)
-                               vmw_bo_dirty_release(res->backup);
-                       vmw_bo_unreference(&res->backup);
+                               vmw_bo_dirty_release(res->guest_memory_bo);
+                       vmw_bo_unreference(&res->guest_memory_bo);
                }
 
-               if (new_backup) {
-                       res->backup = vmw_bo_reference(new_backup);
+               if (new_guest_memory_bo) {
+                       res->guest_memory_bo = vmw_bo_reference(new_guest_memory_bo);
 
                        /*
                         * The validation code should already have added a
                         * dirty tracker here.
                         */
-                       WARN_ON(res->coherent && !new_backup->dirty);
+                       WARN_ON(res->coherent && !new_guest_memory_bo->dirty);
 
                        vmw_resource_mob_attach(res);
                } else {
-                       res->backup = NULL;
+                       res->guest_memory_bo = NULL;
                }
-       } else if (switch_backup && res->coherent) {
-               vmw_bo_dirty_release(res->backup);
+       } else if (switch_guest_memory && res->coherent) {
+               vmw_bo_dirty_release(res->guest_memory_bo);
        }
 
-       if (switch_backup)
-               res->backup_offset = new_backup_offset;
+       if (switch_guest_memory)
+               res->guest_memory_offset = new_guest_memory_offset;
 
        if (dirty_set)
                res->res_dirty = dirty;
@@ -507,32 +512,32 @@ vmw_resource_check_buffer(struct ww_acquire_ctx *ticket,
 {
        struct ttm_operation_ctx ctx = { true, false };
        struct list_head val_list;
-       bool backup_dirty = false;
+       bool guest_memory_dirty = false;
        int ret;
 
-       if (unlikely(res->backup == NULL)) {
+       if (unlikely(!res->guest_memory_bo)) {
                ret = vmw_resource_buf_alloc(res, interruptible);
                if (unlikely(ret != 0))
                        return ret;
        }
 
        INIT_LIST_HEAD(&val_list);
-       ttm_bo_get(&res->backup->base);
-       val_buf->bo = &res->backup->base;
+       ttm_bo_get(&res->guest_memory_bo->tbo);
+       val_buf->bo = &res->guest_memory_bo->tbo;
        val_buf->num_shared = 0;
        list_add_tail(&val_buf->head, &val_list);
        ret = ttm_eu_reserve_buffers(ticket, &val_list, interruptible, NULL);
        if (unlikely(ret != 0))
                goto out_no_reserve;
 
-       if (res->func->needs_backup && !vmw_resource_mob_attached(res))
+       if (res->func->needs_guest_memory && !vmw_resource_mob_attached(res))
                return 0;
 
-       backup_dirty = res->backup_dirty;
-       vmw_bo_placement_set(res->backup, res->func->domain,
+       guest_memory_dirty = res->guest_memory_dirty;
+       vmw_bo_placement_set(res->guest_memory_bo, res->func->domain,
                             res->func->busy_domain);
-       ret = ttm_bo_validate(&res->backup->base,
-                             &res->backup->placement,
+       ret = ttm_bo_validate(&res->guest_memory_bo->tbo,
+                             &res->guest_memory_bo->placement,
                              &ctx);
 
        if (unlikely(ret != 0))
@@ -545,8 +550,8 @@ out_no_validate:
 out_no_reserve:
        ttm_bo_put(val_buf->bo);
        val_buf->bo = NULL;
-       if (backup_dirty)
-               vmw_bo_unreference(&res->backup);
+       if (guest_memory_dirty)
+               vmw_bo_unreference(&res->guest_memory_bo);
 
        return ret;
 }
@@ -557,12 +562,13 @@ out_no_reserve:
  * @res:            The resource to reserve.
  *
  * This function takes the resource off the LRU list and make sure
- * a backup buffer is present for guest-backed resources. However,
- * the buffer may not be bound to the resource at this point.
+ * a guest memory buffer is present for guest-backed resources.
+ * However, the buffer may not be bound to the resource at this
+ * point.
  *
  */
 int vmw_resource_reserve(struct vmw_resource *res, bool interruptible,
-                        bool no_backup)
+                        bool no_guest_memory)
 {
        struct vmw_private *dev_priv = res->dev_priv;
        int ret;
@@ -571,13 +577,13 @@ int vmw_resource_reserve(struct vmw_resource *res, bool interruptible,
        list_del_init(&res->lru_head);
        spin_unlock(&dev_priv->resource_lock);
 
-       if (res->func->needs_backup && res->backup == NULL &&
-           !no_backup) {
+       if (res->func->needs_guest_memory && !res->guest_memory_bo &&
+           !no_guest_memory) {
                ret = vmw_resource_buf_alloc(res, interruptible);
                if (unlikely(ret != 0)) {
-                       DRM_ERROR("Failed to allocate a backup buffer "
+                       DRM_ERROR("Failed to allocate a guest memory buffer "
                                  "of size %lu. bytes\n",
-                                 (unsigned long) res->backup_size);
+                                 (unsigned long) res->guest_memory_size);
                        return ret;
                }
        }
@@ -587,10 +593,10 @@ int vmw_resource_reserve(struct vmw_resource *res, bool interruptible,
 
 /**
  * vmw_resource_backoff_reservation - Unreserve and unreference a
- *                                    backup buffer
+ *                                    guest memory buffer
  *.
  * @ticket:         The ww acquire ctx used for reservation.
- * @val_buf:        Backup buffer information.
+ * @val_buf:        Guest memory buffer information.
  */
 static void
 vmw_resource_backoff_reservation(struct ww_acquire_ctx *ticket,
@@ -632,14 +638,14 @@ static int vmw_resource_do_evict(struct ww_acquire_ctx *ticket,
                return ret;
 
        if (unlikely(func->unbind != NULL &&
-                    (!func->needs_backup || vmw_resource_mob_attached(res)))) {
+                    (!func->needs_guest_memory || vmw_resource_mob_attached(res)))) {
                ret = func->unbind(res, res->res_dirty, &val_buf);
                if (unlikely(ret != 0))
                        goto out_no_unbind;
                vmw_resource_mob_detach(res);
        }
        ret = func->destroy(res);
-       res->backup_dirty = true;
+       res->guest_memory_dirty = true;
        res->res_dirty = false;
 out_no_unbind:
        vmw_resource_backoff_reservation(ticket, &val_buf);
@@ -678,8 +684,8 @@ int vmw_resource_validate(struct vmw_resource *res, bool intr,
 
        val_buf.bo = NULL;
        val_buf.num_shared = 0;
-       if (res->backup)
-               val_buf.bo = &res->backup->base;
+       if (res->guest_memory_bo)
+               val_buf.bo = &res->guest_memory_bo->tbo;
        do {
                ret = vmw_resource_do_validate(res, &val_buf, dirtying);
                if (likely(ret != -EBUSY))
@@ -719,9 +725,9 @@ int vmw_resource_validate(struct vmw_resource *res, bool intr,
 
        if (unlikely(ret != 0))
                goto out_no_validate;
-       else if (!res->func->needs_backup && res->backup) {
+       else if (!res->func->needs_guest_memory && res->guest_memory_bo) {
                WARN_ON_ONCE(vmw_resource_mob_attached(res));
-               vmw_bo_unreference(&res->backup);
+               vmw_bo_unreference(&res->guest_memory_bo);
        }
 
        return 0;
@@ -745,11 +751,11 @@ out_no_validate:
 void vmw_resource_unbind_list(struct vmw_bo *vbo)
 {
        struct ttm_validate_buffer val_buf = {
-               .bo = &vbo->base,
+               .bo = &vbo->tbo,
                .num_shared = 0
        };
 
-       dma_resv_assert_held(vbo->base.base.resv);
+       dma_resv_assert_held(vbo->tbo.base.resv);
        while (!RB_EMPTY_ROOT(&vbo->res_tree)) {
                struct rb_node *node = vbo->res_tree.rb_node;
                struct vmw_resource *res =
@@ -758,12 +764,12 @@ void vmw_resource_unbind_list(struct vmw_bo *vbo)
                if (!WARN_ON_ONCE(!res->func->unbind))
                        (void) res->func->unbind(res, res->res_dirty, &val_buf);
 
-               res->backup_dirty = true;
+               res->guest_memory_size = true;
                res->res_dirty = false;
                vmw_resource_mob_detach(res);
        }
 
-       (void) ttm_bo_wait(&vbo->base, false, false);
+       (void) ttm_bo_wait(&vbo->tbo, false, false);
 }
 
 
@@ -826,9 +832,7 @@ void vmw_query_move_notify(struct ttm_buffer_object *bo,
 {
        struct vmw_bo *dx_query_mob;
        struct ttm_device *bdev = bo->bdev;
-       struct vmw_private *dev_priv;
-
-       dev_priv = container_of(bdev, struct vmw_private, bdev);
+       struct vmw_private *dev_priv = vmw_priv_from_ttm(bdev);
 
        mutex_lock(&dev_priv->binding_mutex);
 
@@ -837,7 +841,7 @@ void vmw_query_move_notify(struct ttm_buffer_object *bo,
            old_mem->mem_type == VMW_PL_MOB) {
                struct vmw_fence_obj *fence;
 
-               dx_query_mob = container_of(bo, struct vmw_bo, base);
+               dx_query_mob = to_vmw_bo(&bo->base);
                if (!dx_query_mob || !dx_query_mob->dx_query_ctx) {
                        mutex_unlock(&dev_priv->binding_mutex);
                        return;
@@ -865,7 +869,7 @@ void vmw_query_move_notify(struct ttm_buffer_object *bo,
  */
 bool vmw_resource_needs_backup(const struct vmw_resource *res)
 {
-       return res->func->needs_backup;
+       return res->func->needs_guest_memory;
 }
 
 /**
@@ -963,22 +967,22 @@ int vmw_resource_pin(struct vmw_resource *res, bool interruptible)
        if (res->pin_count == 0) {
                struct vmw_bo *vbo = NULL;
 
-               if (res->backup) {
-                       vbo = res->backup;
+               if (res->guest_memory_bo) {
+                       vbo = res->guest_memory_bo;
 
-                       ret = ttm_bo_reserve(&vbo->base, interruptible, false, NULL);
+                       ret = ttm_bo_reserve(&vbo->tbo, interruptible, false, NULL);
                        if (ret)
                                goto out_no_validate;
-                       if (!vbo->base.pin_count) {
+                       if (!vbo->tbo.pin_count) {
                                vmw_bo_placement_set(vbo,
                                                     res->func->domain,
                                                     res->func->busy_domain);
                                ret = ttm_bo_validate
-                                       (&vbo->base,
+                                       (&vbo->tbo,
                                         &vbo->placement,
                                         &ctx);
                                if (ret) {
-                                       ttm_bo_unreserve(&vbo->base);
+                                       ttm_bo_unreserve(&vbo->tbo);
                                        goto out_no_validate;
                                }
                        }
@@ -988,7 +992,7 @@ int vmw_resource_pin(struct vmw_resource *res, bool interruptible)
                }
                ret = vmw_resource_validate(res, interruptible, true);
                if (vbo)
-                       ttm_bo_unreserve(&vbo->base);
+                       ttm_bo_unreserve(&vbo->tbo);
                if (ret)
                        goto out_no_validate;
        }
@@ -1021,12 +1025,12 @@ void vmw_resource_unpin(struct vmw_resource *res)
        WARN_ON(ret);
 
        WARN_ON(res->pin_count == 0);
-       if (--res->pin_count == 0 && res->backup) {
-               struct vmw_bo *vbo = res->backup;
+       if (--res->pin_count == 0 && res->guest_memory_bo) {
+               struct vmw_bo *vbo = res->guest_memory_bo;
 
-               (void) ttm_bo_reserve(&vbo->base, false, false, NULL);
+               (void) ttm_bo_reserve(&vbo->tbo, false, false, NULL);
                vmw_bo_pin_reserved(vbo, false);
-               ttm_bo_unreserve(&vbo->base);
+               ttm_bo_unreserve(&vbo->tbo);
        }
 
        vmw_resource_unreserve(res, false, false, false, NULL, 0UL);
@@ -1084,9 +1088,9 @@ int vmw_resources_clean(struct vmw_bo *vbo, pgoff_t start,
                struct vmw_resource *cur_res =
                        container_of(cur, struct vmw_resource, mob_node);
 
-               if (cur_res->backup_offset >= res_end) {
+               if (cur_res->guest_memory_offset >= res_end) {
                        cur = cur->rb_left;
-               } else if (cur_res->backup_offset + cur_res->backup_size <=
+               } else if (cur_res->guest_memory_offset + cur_res->guest_memory_size <=
                           res_start) {
                        cur = cur->rb_right;
                } else {
@@ -1097,7 +1101,7 @@ int vmw_resources_clean(struct vmw_bo *vbo, pgoff_t start,
        }
 
        /*
-        * In order of increasing backup_offset, clean dirty resources
+        * In order of increasing guest_memory_offset, clean dirty resources
         * intersecting the range.
         */
        while (found) {
@@ -1113,13 +1117,13 @@ int vmw_resources_clean(struct vmw_bo *vbo, pgoff_t start,
 
                        found->res_dirty = false;
                }
-               last_cleaned = found->backup_offset + found->backup_size;
+               last_cleaned = found->guest_memory_offset + found->guest_memory_size;
                cur = rb_next(&found->mob_node);
                if (!cur)
                        break;
 
                found = container_of(cur, struct vmw_resource, mob_node);
-               if (found->backup_offset >= res_end)
+               if (found->guest_memory_offset >= res_end)
                        break;
        }
 
@@ -1128,7 +1132,7 @@ int vmw_resources_clean(struct vmw_bo *vbo, pgoff_t start,
         */
        *num_prefault = 1;
        if (last_cleaned > res_start) {
-               struct ttm_buffer_object *bo = &vbo->base;
+               struct ttm_buffer_object *bo = &vbo->tbo;
 
                *num_prefault = __KERNEL_DIV_ROUND_UP(last_cleaned - res_start,
                                                      PAGE_SIZE);
index 2c24e0929faab8da08a21d339cedd722ecd4346a..aa7cbd396bea89415321c3cff93702884d2ab307 100644 (file)
@@ -58,10 +58,11 @@ struct vmw_user_resource_conv {
  * struct vmw_res_func - members and functions common for a resource type
  *
  * @res_type:          Enum that identifies the lru list to use for eviction.
- * @needs_backup:      Whether the resource is guest-backed and needs
+ * @needs_guest_memory:Whether the resource is guest-backed and needs
  *                     persistent buffer storage.
  * @type_name:         String that identifies the resource type.
- * @backup_placement:  TTM placement for backup buffers.
+ * @domain:            TTM placement for guest memory buffers.
+ * @busy_domain:       TTM busy placement for guest memory buffers.
  * @may_evict          Whether the resource may be evicted.
  * @create:            Create a hardware resource.
  * @destroy:           Destroy a hardware resource.
@@ -81,7 +82,7 @@ struct vmw_user_resource_conv {
  */
 struct vmw_res_func {
        enum vmw_res_type res_type;
-       bool needs_backup;
+       bool needs_guest_memory;
        const char *type_name;
        u32 domain;
        u32 busy_domain;
index e9d03ef9815439dc451aec6cc29453acc23ef40b..556a403b7eb56a611a92fc995d1d5b99a2e9a7c8 100644 (file)
@@ -149,7 +149,7 @@ static int vmw_sou_fifo_create(struct vmw_private *dev_priv,
        sou->base.set_gui_y = cmd->obj.root.y;
 
        /* Ok to assume that buffer is pinned in vram */
-       vmw_bo_get_guest_ptr(&sou->buffer->base, &cmd->obj.backingStore.ptr);
+       vmw_bo_get_guest_ptr(&sou->buffer->tbo, &cmd->obj.backingStore.ptr);
        cmd->obj.backingStore.pitch = mode->hdisplay * 4;
 
        vmw_cmd_commit(dev_priv, fifo_size);
@@ -410,9 +410,13 @@ vmw_sou_primary_plane_prepare_fb(struct drm_plane *plane,
        struct drm_crtc *crtc = plane->state->crtc ?: new_state->crtc;
        struct vmw_plane_state *vps = vmw_plane_state_to_vps(new_state);
        struct vmw_private *dev_priv;
-       size_t size;
        int ret;
-
+       struct vmw_bo_params bo_params = {
+               .domain = VMW_BO_DOMAIN_VRAM,
+               .busy_domain = VMW_BO_DOMAIN_VRAM,
+               .bo_type = ttm_bo_type_device,
+               .pin = true
+       };
 
        if (!new_fb) {
                vmw_bo_unreference(&vps->bo);
@@ -421,11 +425,11 @@ vmw_sou_primary_plane_prepare_fb(struct drm_plane *plane,
                return 0;
        }
 
-       size = new_state->crtc_w * new_state->crtc_h * 4;
+       bo_params.size = new_state->crtc_w * new_state->crtc_h * 4;
        dev_priv = vmw_priv(crtc->dev);
 
        if (vps->bo) {
-               if (vps->bo_size == size) {
+               if (vps->bo_size == bo_params.size) {
                        /*
                         * Note that this might temporarily up the pin-count
                         * to 2, until cleanup_fb() is called.
@@ -444,17 +448,12 @@ vmw_sou_primary_plane_prepare_fb(struct drm_plane *plane,
         * resume the overlays, this is preferred to failing to alloc.
         */
        vmw_overlay_pause_all(dev_priv);
-       ret = vmw_bo_create(dev_priv, size,
-                           VMW_BO_DOMAIN_VRAM,
-                           VMW_BO_DOMAIN_VRAM,
-                           false, true, &vps->bo);
+       ret = vmw_bo_create(dev_priv, &bo_params, &vps->bo);
        vmw_overlay_resume_all(dev_priv);
-       if (ret) {
-               vps->bo = NULL; /* vmw_bo_init frees on error */
+       if (ret)
                return ret;
-       }
 
-       vps->bo_size = size;
+       vps->bo_size = bo_params.size;
 
        /*
         * TTM already thinks the buffer is pinned, but make sure the
@@ -491,7 +490,7 @@ static uint32_t vmw_sou_bo_define_gmrfb(struct vmw_du_update_plane *update,
        gmr->body.format.colorDepth = depth;
        gmr->body.format.reserved = 0;
        gmr->body.bytesPerLine = update->vfb->base.pitches[0];
-       vmw_bo_get_guest_ptr(&vfbbo->buffer->base, &gmr->body.ptr);
+       vmw_bo_get_guest_ptr(&vfbbo->buffer->tbo, &gmr->body.ptr);
 
        return sizeof(*gmr);
 }
@@ -973,7 +972,7 @@ static int do_bo_define_gmrfb(struct vmw_private *dev_priv,
        cmd->body.format.reserved = 0;
        cmd->body.bytesPerLine = framebuffer->base.pitches[0];
        /* Buffer is reserved in vram or GMR */
-       vmw_bo_get_guest_ptr(&buf->base, &cmd->body.ptr);
+       vmw_bo_get_guest_ptr(&buf->tbo, &cmd->body.ptr);
        vmw_cmd_commit(dev_priv, sizeof(*cmd));
 
        return 0;
index 9920c103bffb76dfac7d9c0f4d8b5209026cef2d..6b8e984695ed0dc7466c1b658b511239139a484d 100644 (file)
@@ -89,7 +89,7 @@ const struct vmw_user_resource_conv *user_shader_converter =
 
 static const struct vmw_res_func vmw_gb_shader_func = {
        .res_type = vmw_res_shader,
-       .needs_backup = true,
+       .needs_guest_memory = true,
        .may_evict = true,
        .prio = 3,
        .dirty_prio = 3,
@@ -104,7 +104,7 @@ static const struct vmw_res_func vmw_gb_shader_func = {
 
 static const struct vmw_res_func vmw_dx_shader_func = {
        .res_type = vmw_res_shader,
-       .needs_backup = true,
+       .needs_guest_memory = true,
        .may_evict = true,
        .prio = 3,
        .dirty_prio = 3,
@@ -178,10 +178,10 @@ static int vmw_gb_shader_init(struct vmw_private *dev_priv,
                return ret;
        }
 
-       res->backup_size = size;
+       res->guest_memory_size = size;
        if (byte_code) {
-               res->backup = vmw_bo_reference(byte_code);
-               res->backup_offset = offset;
+               res->guest_memory_bo = vmw_bo_reference(byte_code);
+               res->guest_memory_offset = offset;
        }
        shader->size = size;
        shader->type = type;
@@ -262,8 +262,8 @@ static int vmw_gb_shader_bind(struct vmw_resource *res,
        cmd->header.size = sizeof(cmd->body);
        cmd->body.shid = res->id;
        cmd->body.mobid = bo->resource->start;
-       cmd->body.offsetInBytes = res->backup_offset;
-       res->backup_dirty = false;
+       cmd->body.offsetInBytes = res->guest_memory_offset;
+       res->guest_memory_dirty = false;
        vmw_cmd_commit(dev_priv, sizeof(*cmd));
 
        return 0;
@@ -280,7 +280,7 @@ static int vmw_gb_shader_unbind(struct vmw_resource *res,
        } *cmd;
        struct vmw_fence_obj *fence;
 
-       BUG_ON(res->backup->base.resource->mem_type != VMW_PL_MOB);
+       BUG_ON(res->guest_memory_bo->tbo.resource->mem_type != VMW_PL_MOB);
 
        cmd = VMW_CMD_RESERVE(dev_priv, sizeof(*cmd));
        if (unlikely(cmd == NULL))
@@ -400,8 +400,8 @@ static int vmw_dx_shader_unscrub(struct vmw_resource *res)
        cmd->header.size = sizeof(cmd->body);
        cmd->body.cid = shader->ctx->id;
        cmd->body.shid = shader->id;
-       cmd->body.mobid = res->backup->base.resource->start;
-       cmd->body.offsetInBytes = res->backup_offset;
+       cmd->body.mobid = res->guest_memory_bo->tbo.resource->start;
+       cmd->body.offsetInBytes = res->guest_memory_offset;
        vmw_cmd_commit(dev_priv, sizeof(*cmd));
 
        vmw_cotable_add_resource(shader->cotable, &shader->cotable_head);
@@ -511,7 +511,7 @@ static int vmw_dx_shader_unbind(struct vmw_resource *res,
        struct vmw_fence_obj *fence;
        int ret;
 
-       BUG_ON(res->backup->base.resource->mem_type != VMW_PL_MOB);
+       BUG_ON(res->guest_memory_bo->tbo.resource->mem_type != VMW_PL_MOB);
 
        mutex_lock(&dev_priv->binding_mutex);
        ret = vmw_dx_shader_scrub(res);
@@ -785,7 +785,7 @@ static int vmw_shader_define(struct drm_device *dev, struct drm_file *file_priv,
                        return ret;
                }
 
-               if ((u64)buffer->base.base.size < (u64)size + (u64)offset) {
+               if ((u64)buffer->tbo.base.size < (u64)size + (u64)offset) {
                        VMW_DEBUG_USER("Illegal buffer- or shader size.\n");
                        ret = -EINVAL;
                        goto out_bad_arg;
@@ -891,25 +891,29 @@ int vmw_compat_shader_add(struct vmw_private *dev_priv,
        bool is_iomem;
        int ret;
        struct vmw_resource *res;
+       struct vmw_bo_params bo_params = {
+               .domain = VMW_BO_DOMAIN_SYS,
+               .busy_domain = VMW_BO_DOMAIN_SYS,
+               .bo_type = ttm_bo_type_device,
+               .size = size,
+               .pin = true
+       };
 
        if (!vmw_shader_id_ok(user_key, shader_type))
                return -EINVAL;
 
-       ret = vmw_bo_create(dev_priv, size,
-                           VMW_BO_DOMAIN_SYS,
-                           VMW_BO_DOMAIN_SYS,
-                           true, true, &buf);
+       ret = vmw_bo_create(dev_priv, &bo_params, &buf);
        if (unlikely(ret != 0))
                goto out;
 
-       ret = ttm_bo_reserve(&buf->base, false, true, NULL);
+       ret = ttm_bo_reserve(&buf->tbo, false, true, NULL);
        if (unlikely(ret != 0))
                goto no_reserve;
 
        /* Map and copy shader bytecode. */
-       ret = ttm_bo_kmap(&buf->base, 0, PFN_UP(size), &map);
+       ret = ttm_bo_kmap(&buf->tbo, 0, PFN_UP(size), &map);
        if (unlikely(ret != 0)) {
-               ttm_bo_unreserve(&buf->base);
+               ttm_bo_unreserve(&buf->tbo);
                goto no_reserve;
        }
 
@@ -917,12 +921,9 @@ int vmw_compat_shader_add(struct vmw_private *dev_priv,
        WARN_ON(is_iomem);
 
        ttm_bo_kunmap(&map);
-       vmw_bo_placement_set(buf,
-                            VMW_BO_DOMAIN_SYS,
-                            VMW_BO_DOMAIN_SYS);
-       ret = ttm_bo_validate(&buf->base, &buf->placement, &ctx);
+       ret = ttm_bo_validate(&buf->tbo, &buf->placement, &ctx);
        WARN_ON(ret != 0);
-       ttm_bo_unreserve(&buf->base);
+       ttm_bo_unreserve(&buf->tbo);
 
        res = vmw_shader_alloc(dev_priv, buf, size, 0, shader_type);
        if (unlikely(ret != 0))
index 603175b8c97e075d8507885fe4dcefe960acef4f..5af4db6d1f1881228247cfb285278a0e4b67877b 100644 (file)
@@ -82,7 +82,7 @@ static void vmw_view_commit_notify(struct vmw_resource *res,
 
 static const struct vmw_res_func vmw_view_func = {
        .res_type = vmw_res_view,
-       .needs_backup = false,
+       .needs_guest_memory = false,
        .may_evict = false,
        .type_name = "DX view",
        .domain = VMW_BO_DOMAIN_SYS,
index 0df86402e9ceac88a52f90c404c2b38d3c2beefa..d79a6eccfaa440e3ed5cc3d4909f4322a6b0340c 100644 (file)
@@ -505,11 +505,11 @@ static void vmw_stdu_bo_cpu_commit(struct vmw_kms_dirty *dirty)
 
        /* Assume we are blitting from Guest (bo) to Host (display_srf) */
        src_pitch = stdu->display_srf->metadata.base_size.width * stdu->cpp;
-       src_bo = &stdu->display_srf->res.backup->base;
+       src_bo = &stdu->display_srf->res.guest_memory_bo->tbo;
        src_offset = ddirty->top * dst_pitch + ddirty->left * stdu->cpp;
 
        dst_pitch = ddirty->pitch;
-       dst_bo = &ddirty->buf->base;
+       dst_bo = &ddirty->buf->tbo;
        dst_offset = ddirty->fb_top * src_pitch + ddirty->fb_left * stdu->cpp;
 
        (void) vmw_bo_cpu_blit(dst_bo, dst_offset, dst_pitch,
@@ -1080,11 +1080,11 @@ vmw_stdu_bo_populate_update_cpu(struct vmw_du_update_plane  *update, void *cmd,
 
        diff.cpp = stdu->cpp;
 
-       dst_bo = &stdu->display_srf->res.backup->base;
+       dst_bo = &stdu->display_srf->res.guest_memory_bo->tbo;
        dst_pitch = stdu->display_srf->metadata.base_size.width * stdu->cpp;
        dst_offset = bb->y1 * dst_pitch + bb->x1 * stdu->cpp;
 
-       src_bo = &vfbbo->buffer->base;
+       src_bo = &vfbbo->buffer->tbo;
        src_pitch = update->vfb->base.pitches[0];
        src_offset = bo_update->fb_top * src_pitch + bo_update->fb_left *
                stdu->cpp;
index b35ac195285db30d101081194582f80c8a42faaf..edcc4065903850ee60db1b710de29c2c44d19226 100644 (file)
@@ -63,7 +63,7 @@ static void vmw_dx_streamoutput_commit_notify(struct vmw_resource *res,
 
 static const struct vmw_res_func vmw_dx_streamoutput_func = {
        .res_type = vmw_res_streamoutput,
-       .needs_backup = true,
+       .needs_guest_memory = true,
        .may_evict = false,
        .type_name = "DX streamoutput",
        .domain = VMW_BO_DOMAIN_MOB,
@@ -106,8 +106,8 @@ static int vmw_dx_streamoutput_unscrub(struct vmw_resource *res)
        cmd->header.id = SVGA_3D_CMD_DX_BIND_STREAMOUTPUT;
        cmd->header.size = sizeof(cmd->body);
        cmd->body.soid = so->id;
-       cmd->body.mobid = res->backup->base.resource->start;
-       cmd->body.offsetInBytes = res->backup_offset;
+       cmd->body.mobid = res->guest_memory_bo->tbo.resource->start;
+       cmd->body.offsetInBytes = res->guest_memory_offset;
        cmd->body.sizeInBytes = so->size;
        vmw_cmd_commit(dev_priv, sizeof(*cmd));
 
@@ -197,7 +197,7 @@ static int vmw_dx_streamoutput_unbind(struct vmw_resource *res, bool readback,
        struct vmw_fence_obj *fence;
        int ret;
 
-       if (WARN_ON(res->backup->base.resource->mem_type != VMW_PL_MOB))
+       if (WARN_ON(res->guest_memory_bo->tbo.resource->mem_type != VMW_PL_MOB))
                return -EINVAL;
 
        mutex_lock(&dev_priv->binding_mutex);
index 9c6a691b005ee5f3aa68fa814ba87ea355f9c72b..9d4ae9623a00882fdc3129d4c28e3b0a70b2af4f 100644 (file)
@@ -126,7 +126,7 @@ const struct vmw_user_resource_conv *user_surface_converter =
 
 static const struct vmw_res_func vmw_legacy_surface_func = {
        .res_type = vmw_res_surface,
-       .needs_backup = false,
+       .needs_guest_memory = false,
        .may_evict = true,
        .prio = 1,
        .dirty_prio = 1,
@@ -141,7 +141,7 @@ static const struct vmw_res_func vmw_legacy_surface_func = {
 
 static const struct vmw_res_func vmw_gb_surface_func = {
        .res_type = vmw_res_surface,
-       .needs_backup = true,
+       .needs_guest_memory = true,
        .may_evict = true,
        .prio = 1,
        .dirty_prio = 2,
@@ -382,7 +382,7 @@ static void vmw_hw_surface_destroy(struct vmw_resource *res)
                 */
 
                mutex_lock(&dev_priv->cmdbuf_mutex);
-               dev_priv->used_memory_size -= res->backup_size;
+               dev_priv->used_memory_size -= res->guest_memory_size;
                mutex_unlock(&dev_priv->cmdbuf_mutex);
        }
 }
@@ -412,7 +412,7 @@ static int vmw_legacy_srf_create(struct vmw_resource *res)
                return 0;
 
        srf = vmw_res_to_srf(res);
-       if (unlikely(dev_priv->used_memory_size + res->backup_size >=
+       if (unlikely(dev_priv->used_memory_size + res->guest_memory_size >=
                     dev_priv->memory_size))
                return -EBUSY;
 
@@ -450,7 +450,7 @@ static int vmw_legacy_srf_create(struct vmw_resource *res)
         * Surface memory usage accounting.
         */
 
-       dev_priv->used_memory_size += res->backup_size;
+       dev_priv->used_memory_size += res->guest_memory_size;
        return 0;
 
 out_no_fifo:
@@ -527,7 +527,7 @@ static int vmw_legacy_srf_dma(struct vmw_resource *res,
 static int vmw_legacy_srf_bind(struct vmw_resource *res,
                               struct ttm_validate_buffer *val_buf)
 {
-       if (!res->backup_dirty)
+       if (!res->guest_memory_dirty)
                return 0;
 
        return vmw_legacy_srf_dma(res, val_buf, true);
@@ -586,7 +586,7 @@ static int vmw_legacy_srf_destroy(struct vmw_resource *res)
         * Surface memory usage accounting.
         */
 
-       dev_priv->used_memory_size -= res->backup_size;
+       dev_priv->used_memory_size -= res->guest_memory_size;
 
        /*
         * Release the surface ID.
@@ -686,8 +686,8 @@ static void vmw_user_surface_base_release(struct ttm_base_object **p_base)
            container_of(base, struct vmw_user_surface, prime.base);
        struct vmw_resource *res = &user_srf->srf.res;
 
-       if (base->shareable && res && res->backup)
-               drm_gem_object_put(&res->backup->base.base);
+       if (base->shareable && res && res->guest_memory_bo)
+               drm_gem_object_put(&res->guest_memory_bo->tbo.base);
 
        *p_base = NULL;
        vmw_resource_unreference(&res);
@@ -815,7 +815,7 @@ int vmw_surface_define_ioctl(struct drm_device *dev, void *data,
                        ++cur_size;
                }
        }
-       res->backup_size = cur_bo_offset;
+       res->guest_memory_size = cur_bo_offset;
        if (metadata->scanout &&
            metadata->num_sizes == 1 &&
            metadata->sizes[0].width == VMW_CURSOR_SNOOP_WIDTH &&
@@ -859,19 +859,19 @@ int vmw_surface_define_ioctl(struct drm_device *dev, void *data,
 
                ret = vmw_gem_object_create_with_handle(dev_priv,
                                                        file_priv,
-                                                       res->backup_size,
+                                                       res->guest_memory_size,
                                                        &backup_handle,
-                                                       &res->backup);
+                                                       &res->guest_memory_bo);
                if (unlikely(ret != 0)) {
                        vmw_resource_unreference(&res);
                        goto out_unlock;
                }
-               vmw_bo_reference(res->backup);
-               drm_gem_object_get(&res->backup->base.base);
+               vmw_bo_reference(res->guest_memory_bo);
+               drm_gem_object_get(&res->guest_memory_bo->tbo.base);
        }
 
        tmp = vmw_resource_reference(&srf->res);
-       ret = ttm_prime_object_init(tfile, res->backup_size, &user_srf->prime,
+       ret = ttm_prime_object_init(tfile, res->guest_memory_size, &user_srf->prime,
                                    req->shareable, VMW_RES_SURFACE,
                                    &vmw_user_surface_base_release);
 
@@ -1185,7 +1185,7 @@ static int vmw_gb_surface_bind(struct vmw_resource *res,
 
        BUG_ON(bo->resource->mem_type != VMW_PL_MOB);
 
-       submit_size = sizeof(*cmd1) + (res->backup_dirty ? sizeof(*cmd2) : 0);
+       submit_size = sizeof(*cmd1) + (res->guest_memory_dirty ? sizeof(*cmd2) : 0);
 
        cmd1 = VMW_CMD_RESERVE(dev_priv, submit_size);
        if (unlikely(!cmd1))
@@ -1195,7 +1195,7 @@ static int vmw_gb_surface_bind(struct vmw_resource *res,
        cmd1->header.size = sizeof(cmd1->body);
        cmd1->body.sid = res->id;
        cmd1->body.mobid = bo->resource->start;
-       if (res->backup_dirty) {
+       if (res->guest_memory_dirty) {
                cmd2 = (void *) &cmd1[1];
                cmd2->header.id = SVGA_3D_CMD_UPDATE_GB_SURFACE;
                cmd2->header.size = sizeof(cmd2->body);
@@ -1203,12 +1203,12 @@ static int vmw_gb_surface_bind(struct vmw_resource *res,
        }
        vmw_cmd_commit(dev_priv, submit_size);
 
-       if (res->backup->dirty && res->backup_dirty) {
+       if (res->guest_memory_bo->dirty && res->guest_memory_dirty) {
                /* We've just made a full upload. Cear dirty regions. */
                vmw_bo_dirty_clear_res(res);
        }
 
-       res->backup_dirty = false;
+       res->guest_memory_dirty = false;
 
        return 0;
 }
@@ -1504,11 +1504,11 @@ vmw_gb_surface_define_internal(struct drm_device *dev,
 
        if (req->base.buffer_handle != SVGA3D_INVALID_ID) {
                ret = vmw_user_bo_lookup(file_priv, req->base.buffer_handle,
-                                        &res->backup);
+                                        &res->guest_memory_bo);
                if (ret == 0) {
-                       if (res->backup->base.base.size < res->backup_size) {
+                       if (res->guest_memory_bo->tbo.base.size < res->guest_memory_size) {
                                VMW_DEBUG_USER("Surface backup buffer too small.\n");
-                               vmw_bo_unreference(&res->backup);
+                               vmw_bo_unreference(&res->guest_memory_bo);
                                ret = -EINVAL;
                                goto out_unlock;
                        } else {
@@ -1519,11 +1519,11 @@ vmw_gb_surface_define_internal(struct drm_device *dev,
                   (drm_vmw_surface_flag_create_buffer |
                    drm_vmw_surface_flag_coherent)) {
                ret = vmw_gem_object_create_with_handle(dev_priv, file_priv,
-                                                       res->backup_size,
+                                                       res->guest_memory_size,
                                                        &backup_handle,
-                                                       &res->backup);
+                                                       &res->guest_memory_bo);
                if (ret == 0)
-                       vmw_bo_reference(res->backup);
+                       vmw_bo_reference(res->guest_memory_bo);
        }
 
        if (unlikely(ret != 0)) {
@@ -1532,9 +1532,9 @@ vmw_gb_surface_define_internal(struct drm_device *dev,
        }
 
        if (req->base.drm_surface_flags & drm_vmw_surface_flag_coherent) {
-               struct vmw_bo *backup = res->backup;
+               struct vmw_bo *backup = res->guest_memory_bo;
 
-               ttm_bo_reserve(&backup->base, false, false, NULL);
+               ttm_bo_reserve(&backup->tbo, false, false, NULL);
                if (!res->func->dirty_alloc)
                        ret = -EINVAL;
                if (!ret)
@@ -1543,7 +1543,7 @@ vmw_gb_surface_define_internal(struct drm_device *dev,
                        res->coherent = true;
                        ret = res->func->dirty_alloc(res);
                }
-               ttm_bo_unreserve(&backup->base);
+               ttm_bo_unreserve(&backup->tbo);
                if (ret) {
                        vmw_resource_unreference(&res);
                        goto out_unlock;
@@ -1552,7 +1552,7 @@ vmw_gb_surface_define_internal(struct drm_device *dev,
        }
 
        tmp = vmw_resource_reference(res);
-       ret = ttm_prime_object_init(tfile, res->backup_size, &user_srf->prime,
+       ret = ttm_prime_object_init(tfile, res->guest_memory_size, &user_srf->prime,
                                    req->base.drm_surface_flags &
                                    drm_vmw_surface_flag_shareable,
                                    VMW_RES_SURFACE,
@@ -1565,14 +1565,14 @@ vmw_gb_surface_define_internal(struct drm_device *dev,
        }
 
        rep->handle      = user_srf->prime.base.handle;
-       rep->backup_size = res->backup_size;
-       if (res->backup) {
+       rep->backup_size = res->guest_memory_size;
+       if (res->guest_memory_bo) {
                rep->buffer_map_handle =
-                       drm_vma_node_offset_addr(&res->backup->base.base.vma_node);
-               rep->buffer_size = res->backup->base.base.size;
+                       drm_vma_node_offset_addr(&res->guest_memory_bo->tbo.base.vma_node);
+               rep->buffer_size = res->guest_memory_bo->tbo.base.size;
                rep->buffer_handle = backup_handle;
                if (user_srf->prime.base.shareable)
-                       drm_gem_object_get(&res->backup->base.base);
+                       drm_gem_object_get(&res->guest_memory_bo->tbo.base);
        } else {
                rep->buffer_map_handle = 0;
                rep->buffer_size = 0;
@@ -1614,14 +1614,14 @@ vmw_gb_surface_reference_internal(struct drm_device *dev,
 
        user_srf = container_of(base, struct vmw_user_surface, prime.base);
        srf = &user_srf->srf;
-       if (!srf->res.backup) {
+       if (!srf->res.guest_memory_bo) {
                DRM_ERROR("Shared GB surface is missing a backup buffer.\n");
                goto out_bad_resource;
        }
        metadata = &srf->metadata;
 
        mutex_lock(&dev_priv->cmdbuf_mutex); /* Protect res->backup */
-       ret = drm_gem_handle_create(file_priv, &srf->res.backup->base.base,
+       ret = drm_gem_handle_create(file_priv, &srf->res.guest_memory_bo->tbo.base,
                                    &backup_handle);
        mutex_unlock(&dev_priv->cmdbuf_mutex);
        if (ret != 0) {
@@ -1640,11 +1640,11 @@ vmw_gb_surface_reference_internal(struct drm_device *dev,
        rep->creq.base.buffer_handle = backup_handle;
        rep->creq.base.base_size = metadata->base_size;
        rep->crep.handle = user_srf->prime.base.handle;
-       rep->crep.backup_size = srf->res.backup_size;
+       rep->crep.backup_size = srf->res.guest_memory_size;
        rep->crep.buffer_handle = backup_handle;
        rep->crep.buffer_map_handle =
-               drm_vma_node_offset_addr(&srf->res.backup->base.base.vma_node);
-       rep->crep.buffer_size = srf->res.backup->base.base.size;
+               drm_vma_node_offset_addr(&srf->res.guest_memory_bo->tbo.base.vma_node);
+       rep->crep.buffer_size = srf->res.guest_memory_bo->tbo.base.size;
 
        rep->creq.version = drm_vmw_gb_surface_v1;
        rep->creq.svga3d_flags_upper_32_bits =
@@ -1743,12 +1743,12 @@ static void vmw_surface_tex_dirty_range_add(struct vmw_resource *res,
 {
        struct vmw_surface_dirty *dirty =
                (struct vmw_surface_dirty *) res->dirty;
-       size_t backup_end = res->backup_offset + res->backup_size;
+       size_t backup_end = res->guest_memory_offset + res->guest_memory_size;
        struct vmw_surface_loc loc1, loc2;
        const struct vmw_surface_cache *cache;
 
-       start = max_t(size_t, start, res->backup_offset) - res->backup_offset;
-       end = min(end, backup_end) - res->backup_offset;
+       start = max_t(size_t, start, res->guest_memory_offset) - res->guest_memory_offset;
+       end = min(end, backup_end) - res->guest_memory_offset;
        cache = &dirty->cache;
        vmw_surface_get_loc(cache, &loc1, start);
        vmw_surface_get_loc(cache, &loc2, end - 1);
@@ -1795,13 +1795,13 @@ static void vmw_surface_buf_dirty_range_add(struct vmw_resource *res,
        struct vmw_surface_dirty *dirty =
                (struct vmw_surface_dirty *) res->dirty;
        const struct vmw_surface_cache *cache = &dirty->cache;
-       size_t backup_end = res->backup_offset + cache->mip_chain_bytes;
+       size_t backup_end = res->guest_memory_offset + cache->mip_chain_bytes;
        SVGA3dBox *box = &dirty->boxes[0];
        u32 box_c2;
 
        box->h = box->d = 1;
-       start = max_t(size_t, start, res->backup_offset) - res->backup_offset;
-       end = min(end, backup_end) - res->backup_offset;
+       start = max_t(size_t, start, res->guest_memory_offset) - res->guest_memory_offset;
+       end = min(end, backup_end) - res->guest_memory_offset;
        box_c2 = box->x + box->w;
        if (box->w == 0 || box->x > start)
                box->x = start;
@@ -1817,8 +1817,8 @@ static void vmw_surface_dirty_range_add(struct vmw_resource *res, size_t start,
 {
        struct vmw_surface *srf = vmw_res_to_srf(res);
 
-       if (WARN_ON(end <= res->backup_offset ||
-                   start >= res->backup_offset + res->backup_size))
+       if (WARN_ON(end <= res->guest_memory_offset ||
+                   start >= res->guest_memory_offset + res->guest_memory_size))
                return;
 
        if (srf->metadata.format == SVGA3D_BUFFER)
@@ -2075,7 +2075,7 @@ int vmw_gb_surface_define(struct vmw_private *dev_priv,
        if (metadata->flags & SVGA3D_SURFACE_MULTISAMPLE)
                sample_count = metadata->multisample_count;
 
-       srf->res.backup_size =
+       srf->res.guest_memory_size =
                vmw_surface_get_serialized_size_extended(
                                metadata->format,
                                metadata->base_size,
@@ -2084,7 +2084,7 @@ int vmw_gb_surface_define(struct vmw_private *dev_priv,
                                sample_count);
 
        if (metadata->flags & SVGA3D_SURFACE_BIND_STREAM_OUTPUT)
-               srf->res.backup_size += sizeof(SVGA3dDXSOState);
+               srf->res.guest_memory_size += sizeof(SVGA3dDXSOState);
 
        /*
         * Don't set SVGA3D_SURFACE_SCREENTARGET flag for a scanout surface with
index 1a61ebf039734ed91ec7d5ce36088df5ee732027..4daebc5b9eb4c310566afa6e9849e42cd53ecbe8 100644 (file)
@@ -50,13 +50,6 @@ static const struct ttm_place gmr_placement_flags = {
        .flags = 0
 };
 
-static const struct ttm_place mob_placement_flags = {
-       .fpfn = 0,
-       .lpfn = 0,
-       .mem_type = VMW_PL_MOB,
-       .flags = 0
-};
-
 struct ttm_placement vmw_vram_placement = {
        .num_placement = 1,
        .placement = &vram_placement_flags,
@@ -78,13 +71,6 @@ static const struct ttm_place vram_gmr_placement_flags[] = {
        }
 };
 
-static const struct ttm_place vmw_sys_placement_flags = {
-       .fpfn = 0,
-       .lpfn = 0,
-       .mem_type = VMW_PL_SYSTEM,
-       .flags = 0
-};
-
 struct ttm_placement vmw_vram_gmr_placement = {
        .num_placement = 2,
        .placement = vram_gmr_placement_flags,
@@ -92,13 +78,6 @@ struct ttm_placement vmw_vram_gmr_placement = {
        .busy_placement = &gmr_placement_flags
 };
 
-struct ttm_placement vmw_vram_sys_placement = {
-       .num_placement = 1,
-       .placement = &vram_placement_flags,
-       .num_busy_placement = 1,
-       .busy_placement = &sys_placement_flags
-};
-
 struct ttm_placement vmw_sys_placement = {
        .num_placement = 1,
        .placement = &sys_placement_flags,
@@ -106,20 +85,6 @@ struct ttm_placement vmw_sys_placement = {
        .busy_placement = &sys_placement_flags
 };
 
-struct ttm_placement vmw_pt_sys_placement = {
-       .num_placement = 1,
-       .placement = &vmw_sys_placement_flags,
-       .num_busy_placement = 1,
-       .busy_placement = &vmw_sys_placement_flags
-};
-
-struct ttm_placement vmw_mob_placement = {
-       .num_placement = 1,
-       .num_busy_placement = 1,
-       .placement = &mob_placement_flags,
-       .busy_placement = &mob_placement_flags
-};
-
 const size_t vmw_tt_size = sizeof(struct vmw_ttm_tt);
 
 /**
@@ -462,7 +427,7 @@ static struct ttm_tt *vmw_ttm_tt_create(struct ttm_buffer_object *bo,
        if (!vmw_be)
                return NULL;
 
-       vmw_be->dev_priv = container_of(bo->bdev, struct vmw_private, bdev);
+       vmw_be->dev_priv = vmw_priv_from_ttm(bo->bdev);
        vmw_be->mob = NULL;
 
        if (vmw_be->dev_priv->map_mode == vmw_dma_alloc_coherent)
@@ -488,7 +453,7 @@ static void vmw_evict_flags(struct ttm_buffer_object *bo,
 
 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);
+       struct vmw_private *dev_priv = vmw_priv_from_ttm(bdev);
 
        switch (mem->mem_type) {
        case TTM_PL_SYSTEM:
@@ -599,34 +564,39 @@ struct ttm_device_funcs vmw_bo_driver = {
 };
 
 int vmw_bo_create_and_populate(struct vmw_private *dev_priv,
-                              unsigned long bo_size,
-                              struct ttm_buffer_object **bo_p)
+                              size_t bo_size, u32 domain,
+                              struct vmw_bo **bo_p)
 {
        struct ttm_operation_ctx ctx = {
                .interruptible = false,
                .no_wait_gpu = false
        };
-       struct ttm_buffer_object *bo;
+       struct vmw_bo *vbo;
        int ret;
+       struct vmw_bo_params bo_params = {
+               .domain = domain,
+               .busy_domain = domain,
+               .bo_type = ttm_bo_type_kernel,
+               .size = bo_size,
+               .pin = true
+       };
 
-       ret = vmw_bo_create_kernel(dev_priv, bo_size,
-                                  &vmw_pt_sys_placement,
-                                  &bo);
+       ret = vmw_bo_create(dev_priv, &bo_params, &vbo);
        if (unlikely(ret != 0))
                return ret;
 
-       ret = ttm_bo_reserve(bo, false, true, NULL);
+       ret = ttm_bo_reserve(&vbo->tbo, false, true, NULL);
        BUG_ON(ret != 0);
-       ret = vmw_ttm_populate(bo->bdev, bo->ttm, &ctx);
+       ret = vmw_ttm_populate(vbo->tbo.bdev, vbo->tbo.ttm, &ctx);
        if (likely(ret == 0)) {
                struct vmw_ttm_tt *vmw_tt =
-                       container_of(bo->ttm, struct vmw_ttm_tt, dma_ttm);
+                       container_of(vbo->tbo.ttm, struct vmw_ttm_tt, dma_ttm);
                ret = vmw_ttm_map_dma(vmw_tt);
        }
 
-       ttm_bo_unreserve(bo);
+       ttm_bo_unreserve(&vbo->tbo);
 
        if (likely(ret == 0))
-               *bo_p = bo;
+               *bo_p = vbo;
        return ret;
 }
index c968180c97699dc481f13758be253004ac8c4a6e..d140089e53d4169da34834a9e11751974045f1f4 100644 (file)
@@ -81,7 +81,7 @@ static void vmw_stream_set_arg_handle(void *data, u32 handle)
 static const struct vmw_simple_resource_func va_stream_func = {
        .res_func = {
                .res_type = vmw_res_stream,
-               .needs_backup = false,
+               .needs_guest_memory = false,
                .may_evict = false,
                .type_name = "overlay stream",
                .domain = VMW_BO_DOMAIN_SYS,
index 05f0909ff1dd54c21732dac2227d2e9bbc50f1d7..aaacbdcbd742fbdeb9663350f692470af7b4f71f 100644 (file)
@@ -55,13 +55,13 @@ struct vmw_validation_bo_node {
  * @head: List head for the resource validation list.
  * @hash: A hash entry used for the duplicate detection hash table.
  * @res: Reference counted resource pointer.
- * @new_backup: Non ref-counted pointer to new backup buffer to be assigned
- * to a resource.
- * @new_backup_offset: Offset into the new backup mob for resources that can
- * share MOBs.
+ * @new_guest_memory_bo: Non ref-counted pointer to new guest memory buffer
+ * to be assigned to a resource.
+ * @new_guest_memory_offset: Offset into the new backup mob for resources
+ * that can share MOBs.
  * @no_buffer_needed: Kernel does not need to allocate a MOB during validation,
  * the command stream provides a mob bind operation.
- * @switching_backup: The validation process is switching backup MOB.
+ * @switching_guest_memory_bo: The validation process is switching backup MOB.
  * @first_usage: True iff the resource has been seen only once in the current
  * validation batch.
  * @reserved: Whether the resource is currently reserved by this process.
@@ -76,10 +76,10 @@ struct vmw_validation_res_node {
        struct list_head head;
        struct vmwgfx_hash_item hash;
        struct vmw_resource *res;
-       struct vmw_bo *new_backup;
-       unsigned long new_backup_offset;
+       struct vmw_bo *new_guest_memory_bo;
+       unsigned long new_guest_memory_offset;
        u32 no_buffer_needed : 1;
-       u32 switching_backup : 1;
+       u32 switching_guest_memory_bo : 1;
        u32 first_usage : 1;
        u32 reserved : 1;
        u32 dirty : 1;
@@ -193,7 +193,7 @@ vmw_validation_find_bo_dup(struct vmw_validation_context *ctx,
                struct  vmw_validation_bo_node *entry;
 
                list_for_each_entry(entry, &ctx->bo_list, base.head) {
-                       if (entry->base.bo == &vbo->base) {
+                       if (entry->base.bo == &vbo->tbo) {
                                bo_node = entry;
                                break;
                        }
@@ -279,7 +279,7 @@ int vmw_validation_add_bo(struct vmw_validation_context *ctx,
                                bo_node->hash.key);
                }
                val_buf = &bo_node->base;
-               val_buf->bo = ttm_bo_get_unless_zero(&vbo->base);
+               val_buf->bo = ttm_bo_get_unless_zero(&vbo->tbo);
                if (!val_buf->bo)
                        return -ESRCH;
                val_buf->num_shared = 0;
@@ -393,23 +393,23 @@ void vmw_validation_res_set_dirty(struct vmw_validation_context *ctx,
  * the resource.
  * @vbo: The new backup buffer object MOB. This buffer object needs to have
  * already been registered with the validation context.
- * @backup_offset: Offset into the new backup MOB.
+ * @guest_memory_offset: Offset into the new backup MOB.
  */
 void vmw_validation_res_switch_backup(struct vmw_validation_context *ctx,
                                      void *val_private,
                                      struct vmw_bo *vbo,
-                                     unsigned long backup_offset)
+                                     unsigned long guest_memory_offset)
 {
        struct vmw_validation_res_node *val;
 
        val = container_of(val_private, typeof(*val), private);
 
-       val->switching_backup = 1;
+       val->switching_guest_memory_bo = 1;
        if (val->first_usage)
                val->no_buffer_needed = 1;
 
-       val->new_backup = vbo;
-       val->new_backup_offset = backup_offset;
+       val->new_guest_memory_bo = vbo;
+       val->new_guest_memory_offset = guest_memory_offset;
 }
 
 /**
@@ -437,8 +437,8 @@ int vmw_validation_res_reserve(struct vmw_validation_context *ctx,
                        goto out_unreserve;
 
                val->reserved = 1;
-               if (res->backup) {
-                       struct vmw_bo *vbo = res->backup;
+               if (res->guest_memory_bo) {
+                       struct vmw_bo *vbo = res->guest_memory_bo;
 
                        vmw_bo_placement_set(vbo,
                                             res->func->domain,
@@ -448,11 +448,11 @@ int vmw_validation_res_reserve(struct vmw_validation_context *ctx,
                                goto out_unreserve;
                }
 
-               if (val->switching_backup && val->new_backup &&
+               if (val->switching_guest_memory_bo && val->new_guest_memory_bo &&
                    res->coherent) {
                        struct vmw_validation_bo_node *bo_node =
                                vmw_validation_find_bo_dup(ctx,
-                                                          val->new_backup);
+                                                          val->new_guest_memory_bo);
 
                        if (WARN_ON(!bo_node)) {
                                ret = -EINVAL;
@@ -495,9 +495,9 @@ void vmw_validation_res_unreserve(struct vmw_validation_context *ctx,
                                vmw_resource_unreserve(val->res,
                                                       val->dirty_set,
                                                       val->dirty,
-                                                      val->switching_backup,
-                                                      val->new_backup,
-                                                      val->new_backup_offset);
+                                                      val->switching_guest_memory_bo,
+                                                      val->new_guest_memory_bo,
+                                                      val->new_guest_memory_offset);
                }
 }
 
@@ -512,8 +512,7 @@ void vmw_validation_res_unreserve(struct vmw_validation_context *ctx,
 static int vmw_validation_bo_validate_single(struct ttm_buffer_object *bo,
                                             bool interruptible)
 {
-       struct vmw_bo *vbo =
-               container_of(bo, struct vmw_bo, base);
+       struct vmw_bo *vbo = to_vmw_bo(&bo->base);
        struct ttm_operation_ctx ctx = {
                .interruptible = interruptible,
                .no_wait_gpu = false
@@ -523,7 +522,7 @@ static int vmw_validation_bo_validate_single(struct ttm_buffer_object *bo,
        if (atomic_read(&vbo->cpu_writers))
                return -EBUSY;
 
-       if (vbo->base.pin_count > 0)
+       if (vbo->tbo.pin_count > 0)
                return 0;
 
        ret = ttm_bo_validate(bo, &vbo->placement, &ctx);
@@ -554,8 +553,7 @@ int vmw_validation_bo_validate(struct vmw_validation_context *ctx, bool intr)
        int ret;
 
        list_for_each_entry(entry, &ctx->bo_list, base.head) {
-               struct vmw_bo *vbo =
-                       container_of(entry->base.bo, typeof(*vbo), base);
+               struct vmw_bo *vbo = to_vmw_bo(&entry->base.bo->base);
 
                ret = vmw_validation_bo_validate_single(entry->base.bo, intr);
 
@@ -605,7 +603,7 @@ int vmw_validation_res_validate(struct vmw_validation_context *ctx, bool intr)
 
        list_for_each_entry(val, &ctx->resource_list, head) {
                struct vmw_resource *res = val->res;
-               struct vmw_bo *backup = res->backup;
+               struct vmw_bo *backup = res->guest_memory_bo;
 
                ret = vmw_resource_validate(res, intr, val->dirty_set &&
                                            val->dirty);
@@ -616,8 +614,8 @@ int vmw_validation_res_validate(struct vmw_validation_context *ctx, bool intr)
                }
 
                /* Check if the resource switched backup buffer */
-               if (backup && res->backup && (backup != res->backup)) {
-                       struct vmw_bo *vbo = res->backup;
+               if (backup && res->guest_memory_bo && backup != res->guest_memory_bo) {
+                       struct vmw_bo *vbo = res->guest_memory_bo;
 
                        vmw_bo_placement_set(vbo, res->func->domain,
                                             res->func->busy_domain);
@@ -855,9 +853,7 @@ void vmw_validation_bo_backoff(struct vmw_validation_context *ctx)
        list_for_each_entry(entry, &ctx->bo_list, base.head) {
                if (entry->coherent_count) {
                        unsigned int coherent_count = entry->coherent_count;
-                       struct vmw_bo *vbo =
-                               container_of(entry->base.bo, typeof(*vbo),
-                                            base);
+                       struct vmw_bo *vbo = to_vmw_bo(&entry->base.bo->base);
 
                        while (coherent_count--)
                                vmw_bo_dirty_release(vbo);