On 10/5/23 13:51, Thomas Hellström wrote:
Hi,

On 9/28/23 21:16, Danilo Krummrich wrote:
This patch adds an abstraction layer between the drm_gpuva mappings of
NIT: imperative:  s/This patch adds/Add/
a particular drm_gem_object and this GEM object itself. The abstraction
represents a combination of a drm_gem_object and drm_gpuvm. The
drm_gem_object holds a list of drm_gpuvm_bo structures (the structure
representing this abstraction), while each drm_gpuvm_bo contains list of
mappings of this GEM object.

This has multiple advantages:

1) We can use the drm_gpuvm_bo structure to attach it to various lists
    of the drm_gpuvm. This is useful for tracking external and evicted
    objects per VM, which is introduced in subsequent patches.

2) Finding mappings of a certain drm_gem_object mapped in a certain
    drm_gpuvm becomes much cheaper.

3) Drivers can derive and extend the structure to easily represent
    driver specific states of a BO for a certain GPUVM.

The idea of this abstraction was taken from amdgpu, hence the credit for
this idea goes to the developers of amdgpu.

Cc: Christian König <christian.koe...@amd.com>
Signed-off-by: Danilo Krummrich <d...@redhat.com>
---
  drivers/gpu/drm/drm_gpuvm.c            | 334 +++++++++++++++++++++----
  drivers/gpu/drm/nouveau/nouveau_uvmm.c |  64 +++--
  include/drm/drm_gem.h                  |  32 +--
  include/drm/drm_gpuvm.h                | 177 ++++++++++++-
  4 files changed, 523 insertions(+), 84 deletions(-)

diff --git a/drivers/gpu/drm/drm_gpuvm.c b/drivers/gpu/drm/drm_gpuvm.c
index 6368dfdbe9dd..27100423154b 100644
--- a/drivers/gpu/drm/drm_gpuvm.c
+++ b/drivers/gpu/drm/drm_gpuvm.c
@@ -70,6 +70,18 @@
   * &drm_gem_object, such as the &drm_gem_object containing the root page 
table,
   * but it can also be a 'dummy' object, which can be allocated with
   * drm_gpuvm_root_object_alloc().
+ *
+ * In order to connect a struct drm_gpuva its backing &drm_gem_object each
+ * &drm_gem_object maintains a list of &drm_gpuvm_bo structures, and each
+ * &drm_gpuvm_bo contains a list of &&drm_gpuva structures.
+ *
+ * A &drm_gpuvm_bo is an abstraction that represents a combination of a
+ * &drm_gpuvm and a &drm_gem_object. Every such combination should be unique.
+ * This is ensured by the API through drm_gpuvm_bo_obtain() and
+ * drm_gpuvm_bo_obtain_prealloc() which first look into the corresponding
+ * &drm_gem_object list of &drm_gpuvm_bos for an existing instance of this
+ * particular combination. If not existent a new instance is created and linked
+ * to the &drm_gem_object.
   */
  /**
@@ -395,21 +407,28 @@
  /**
   * DOC: Locking
   *
- * Generally, the GPU VA manager does not take care of locking itself, it is
- * the drivers responsibility to take care about locking. Drivers might want to
- * protect the following operations: inserting, removing and iterating
- * &drm_gpuva objects as well as generating all kinds of operations, such as
- * split / merge or prefetch.
- *
- * The GPU VA manager also does not take care of the locking of the backing
- * &drm_gem_object buffers GPU VA lists by itself; drivers are responsible to
- * enforce mutual exclusion using either the GEMs dma_resv lock or 
alternatively
- * a driver specific external lock. For the latter see also
- * drm_gem_gpuva_set_lock().
- *
- * However, the GPU VA manager contains lockdep checks to ensure callers of its
- * API hold the corresponding lock whenever the &drm_gem_objects GPU VA list is
- * accessed by functions such as drm_gpuva_link() or drm_gpuva_unlink().
+ * In terms of managing &drm_gpuva entries DRM GPUVM does not take care of
+ * locking itself, it is the drivers responsibility to take care about locking.
+ * Drivers might want to protect the following operations: inserting, removing
+ * and iterating &drm_gpuva objects as well as generating all kinds of
+ * operations, such as split / merge or prefetch.
+ *
+ * DRM GPUVM also does not take care of the locking of the backing
+ * &drm_gem_object buffers GPU VA lists and &drm_gpuvm_bo abstractions by
+ * itself; drivers are responsible to enforce mutual exclusion using either the
+ * GEMs dma_resv lock or alternatively a driver specific external lock. For the
+ * latter see also drm_gem_gpuva_set_lock().
+ *
+ * However, DRM GPUVM contains lockdep checks to ensure callers of its API hold
+ * the corresponding lock whenever the &drm_gem_objects GPU VA list is accessed
+ * by functions such as drm_gpuva_link() or drm_gpuva_unlink(), but also
+ * drm_gpuvm_bo_obtain() and drm_gpuvm_bo_put().
+ *
+ * The latter is required since on creation and destruction of a &drm_gpuvm_bo
+ * the &drm_gpuvm_bo is attached / removed from the &drm_gem_objects gpuva 
list.
+ * Subsequent calls to drm_gpuvm_bo_obtain() for the same &drm_gpuvm and
+ * &drm_gem_object must be able to observe previous creations and destructions
+ * of &drm_gpuvm_bos in order to keep instances unique.
   */
  /**
@@ -439,6 +458,7 @@
   *    {
   *        struct drm_gpuva_ops *ops;
   *        struct drm_gpuva_op *op
+ *        struct drm_gpuvm_bo *vm_bo;
   *
   *        driver_lock_va_space();
   *        ops = drm_gpuvm_sm_map_ops_create(gpuvm, addr, range,
@@ -446,6 +466,10 @@
   *        if (IS_ERR(ops))
   *            return PTR_ERR(ops);
   *
+ *        vm_bo = drm_gpuvm_bo_obtain(gpuvm, obj);
+ *        if (IS_ERR(vm_bo))
+ *            return PTR_ERR(vm_bo);
+ *
   *        drm_gpuva_for_each_op(op, ops) {
   *            struct drm_gpuva *va;
   *
@@ -458,7 +482,7 @@
   *
   *                driver_vm_map();
   *                drm_gpuva_map(gpuvm, va, &op->map);
- *                drm_gpuva_link(va);
+ *                drm_gpuva_link(va, vm_bo);
   *
   *                break;
   *            case DRM_GPUVA_OP_REMAP: {
@@ -485,11 +509,11 @@
   *                driver_vm_remap();
   *                drm_gpuva_remap(prev, next, &op->remap);
   *
- *                drm_gpuva_unlink(va);
   *                if (prev)
- *                    drm_gpuva_link(prev);
+ *                    drm_gpuva_link(prev, va->vm_bo);
   *                if (next)
- *                    drm_gpuva_link(next);
+ *                    drm_gpuva_link(next, va->vm_bo);
+ *                drm_gpuva_unlink(va);
   *
   *                break;
   *            }
@@ -505,6 +529,7 @@
   *                break;
   *            }
   *        }
+ *        drm_gpuvm_bo_put(vm_bo);
   *        driver_unlock_va_space();
   *
   *        return 0;
@@ -514,6 +539,7 @@
   *
   *    struct driver_context {
   *        struct drm_gpuvm *gpuvm;
+ *        struct drm_gpuvm_bo *vm_bo;
   *        struct drm_gpuva *new_va;
   *        struct drm_gpuva *prev_va;
   *        struct drm_gpuva *next_va;
@@ -534,6 +560,7 @@
   *                  struct drm_gem_object *obj, u64 offset)
   *    {
   *        struct driver_context ctx;
+ *        struct drm_gpuvm_bo *vm_bo;
   *        struct drm_gpuva_ops *ops;
   *        struct drm_gpuva_op *op;
   *        int ret = 0;
@@ -543,16 +570,23 @@
   *        ctx.new_va = kzalloc(sizeof(*ctx.new_va), GFP_KERNEL);
   *        ctx.prev_va = kzalloc(sizeof(*ctx.prev_va), GFP_KERNEL);
   *        ctx.next_va = kzalloc(sizeof(*ctx.next_va), GFP_KERNEL);
- *        if (!ctx.new_va || !ctx.prev_va || !ctx.next_va) {
+ *        ctx.vm_bo = drm_gpuvm_bo_create(gpuvm, obj);
+ *        if (!ctx.new_va || !ctx.prev_va || !ctx.next_va || !vm_bo) {
   *            ret = -ENOMEM;
   *            goto out;
   *        }
   *
+ *        // Typically protected with a driver specific GEM gpuva lock
+ *        // used in the fence signaling path for drm_gpuva_link() and
+ *        // drm_gpuva_unlink(), hence pre-allocate.
+ *        ctx.vm_bo = drm_gpuvm_bo_obtain_prealloc(ctx.vm_bo);
+ *
   *        driver_lock_va_space();
   *        ret = drm_gpuvm_sm_map(gpuvm, &ctx, addr, range, obj, offset);
   *        driver_unlock_va_space();
   *
   *    out:
+ *        drm_gpuvm_bo_put(ctx.vm_bo);
   *        kfree(ctx.new_va);
   *        kfree(ctx.prev_va);
   *        kfree(ctx.next_va);
@@ -565,7 +599,7 @@
   *
   *        drm_gpuva_map(ctx->vm, ctx->new_va, &op->map);
   *
- *        drm_gpuva_link(ctx->new_va);
+ *        drm_gpuva_link(ctx->new_va, ctx->vm_bo);
   *
   *        // prevent the new GPUVA from being freed in
   *        // driver_mapping_create()
@@ -577,22 +611,23 @@
   *    int driver_gpuva_remap(struct drm_gpuva_op *op, void *__ctx)
   *    {
   *        struct driver_context *ctx = __ctx;
+ *        struct drm_gpuva *va = op->remap.unmap->va;
   *
   *        drm_gpuva_remap(ctx->prev_va, ctx->next_va, &op->remap);
   *
- *        drm_gpuva_unlink(op->remap.unmap->va);
- *        kfree(op->remap.unmap->va);
- *
   *        if (op->remap.prev) {
- *            drm_gpuva_link(ctx->prev_va);
+ *            drm_gpuva_link(ctx->prev_va, va->vm_bo);
   *            ctx->prev_va = NULL;
   *        }
   *
   *        if (op->remap.next) {
- *            drm_gpuva_link(ctx->next_va);
+ *            drm_gpuva_link(ctx->next_va, va->vm_bo);
   *            ctx->next_va = NULL;
   *        }
   *
+ *        drm_gpuva_unlink(va);
+ *        kfree(va);
+ *
   *        return 0;
   *    }
   *
@@ -771,6 +806,194 @@ drm_gpuvm_destroy(struct drm_gpuvm *gpuvm)
  }
  EXPORT_SYMBOL_GPL(drm_gpuvm_destroy);
+/**
+ * drm_gpuvm_bo_create() - create a new instance of struct drm_gpuvm_bo
+ * @gpuvm: The &drm_gpuvm the @obj is mapped in.
+ * @obj: The &drm_gem_object being mapped in the @gpuvm.
+ *
+ * If provided by the driver, this function uses the &drm_gpuvm_ops
+ * vm_bo_alloc() callback to allocate.
+ *
+ * Returns: a pointer to the &drm_gpuvm_bo on success, NULL on failure
+ */
+struct drm_gpuvm_bo *
+drm_gpuvm_bo_create(struct drm_gpuvm *gpuvm,
+            struct drm_gem_object *obj)
+{
+    const struct drm_gpuvm_ops *ops = gpuvm->ops;
+    struct drm_gpuvm_bo *vm_bo;
+
+    if (ops && ops->vm_bo_alloc)
+        vm_bo = ops->vm_bo_alloc();
+    else
+        vm_bo = kzalloc(sizeof(*vm_bo), GFP_KERNEL);
+
+    if (unlikely(!vm_bo))
+        return NULL;
+
+    vm_bo->vm = gpuvm;
+    vm_bo->obj = obj;
+
+    kref_init(&vm_bo->kref);
+    INIT_LIST_HEAD(&vm_bo->list.gpuva);
+    INIT_LIST_HEAD(&vm_bo->list.entry.gem);
+
+    drm_gem_object_get(obj);
+
+    return vm_bo;
+}
+EXPORT_SYMBOL_GPL(drm_gpuvm_bo_create);
+
+static void
+drm_gpuvm_bo_destroy(struct kref *kref)
+{
+    struct drm_gpuvm_bo *vm_bo = container_of(kref, struct drm_gpuvm_bo,
+                          kref);
+    struct drm_gpuvm *gpuvm = vm_bo->vm;
+    const struct drm_gpuvm_ops *ops = gpuvm->ops;
+    struct drm_gem_object *obj = vm_bo->obj;
+    bool lock = !drm_gpuvm_resv_protected(gpuvm);
+
+    drm_gem_gpuva_assert_lock_held(obj);
+    if (!lock)
+        drm_gpuvm_resv_assert_held(gpuvm);
+
+    list_del(&vm_bo->list.entry.gem);
+
+    drm_gem_object_put(obj);
+
+    if (ops && ops->vm_bo_free)
+        ops->vm_bo_free(vm_bo);
+    else
+        kfree(vm_bo);
+}
+
+/**
+ * drm_gpuvm_bo_put() - drop a struct drm_gpuvm_bo reference
+ * @vm_bo: the &drm_gpuvm_bo to release the reference of
+ *
+ * This releases a reference to @vm_bo.
+ *
+ * If the reference count drops to zero, the &gpuvm_bo is destroyed, which
+ * includes removing it from the GEMs gpuva list. Hence, if a call to this
+ * function can potentially let the reference count to zero the caller must
+ * hold the dma-resv or driver specific GEM gpuva lock.
+ */
+void
+drm_gpuvm_bo_put(struct drm_gpuvm_bo *vm_bo)
+{
+    if (vm_bo)
+        kref_put(&vm_bo->kref, drm_gpuvm_bo_destroy);
+}
+EXPORT_SYMBOL_GPL(drm_gpuvm_bo_put);
+
+static struct drm_gpuvm_bo *
+__drm_gpuvm_bo_find(struct drm_gpuvm *gpuvm,
+            struct drm_gem_object *obj)
+{
+    struct drm_gpuvm_bo *vm_bo;
+
+    drm_gem_gpuva_assert_lock_held(obj);
+
+    drm_gem_for_each_gpuvm_bo(vm_bo, obj)
+        if (vm_bo->vm == gpuvm)
+            return vm_bo;
+
+    return NULL;
+}
+
+/**
+ * drm_gpuvm_bo_find() - find the &drm_gpuvm_bo for the given
+ * &drm_gpuvm and &drm_gem_object
+ * @gpuvm: The &drm_gpuvm the @obj is mapped in.
+ * @obj: The &drm_gem_object being mapped in the @gpuvm.
+ *
+ * Find the &drm_gpuvm_bo representing the combination of the given
+ * &drm_gpuvm and &drm_gem_object. If found, increases the reference
+ * count of the &drm_gpuvm_bo accordingly.
+ *
+ * Returns: a pointer to the &drm_gpuvm_bo on success, NULL on failure
+ */
+struct drm_gpuvm_bo *
+drm_gpuvm_bo_find(struct drm_gpuvm *gpuvm,
+          struct drm_gem_object *obj)
+{
+    struct drm_gpuvm_bo *vm_bo = __drm_gpuvm_bo_find(gpuvm, obj);
+
+    return vm_bo ? drm_gpuvm_bo_get(vm_bo) : NULL;
+}
+EXPORT_SYMBOL_GPL(drm_gpuvm_bo_find);
+
+/**
+ * drm_gpuvm_bo_obtain() - obtains and instance of the &drm_gpuvm_bo for the
+ * given &drm_gpuvm and &drm_gem_object
+ * @gpuvm: The &drm_gpuvm the @obj is mapped in.
+ * @obj: The &drm_gem_object being mapped in the @gpuvm.
+ *
+ * Find the &drm_gpuvm_bo representing the combination of the given
+ * &drm_gpuvm and &drm_gem_object. If found, increases the reference
+ * count of the &drm_gpuvm_bo accordingly. If not found, allocates a new
+ * &drm_gpuvm_bo.
+ *
+ * A new &drm_gpuvm_bo is added to the GEMs gpuva list.
+ *
+ * Returns: a pointer to the &drm_gpuvm_bo on success, an ERR_PTR on failure
+ */
+struct drm_gpuvm_bo *
+drm_gpuvm_bo_obtain(struct drm_gpuvm *gpuvm,
+            struct drm_gem_object *obj)
+{
+    struct drm_gpuvm_bo *vm_bo;
+
+    vm_bo = drm_gpuvm_bo_find(gpuvm, obj);
+    if (vm_bo)
+        return vm_bo;
+
+    vm_bo = drm_gpuvm_bo_create(gpuvm, obj);
+    if (!vm_bo)
+        return ERR_PTR(-ENOMEM);
+
+    list_add_tail(&vm_bo->list.entry.gem, &obj->gpuva.list);
+
+    return vm_bo;
+}
+EXPORT_SYMBOL_GPL(drm_gpuvm_bo_obtain);
+
+/**
+ * drm_gpuvm_bo_obtain_prealloc() - obtains and instance of the &drm_gpuvm_bo
+ * for the given &drm_gpuvm and &drm_gem_object
+ * @__vm_bo: A pre-allocated struct drm_gpuvm_bo.
+ *
+ * Find the &drm_gpuvm_bo representing the combination of the given
+ * &drm_gpuvm and &drm_gem_object. If found, increases the reference
+ * count of the found &drm_gpuvm_bo accordingly, while the @__vm_bo reference
+ * count is decreased. If not found @__vm_bo is returned without further
+ * increase of the reference count.
+ *
+ * A new &drm_gpuvm_bo is added to the GEMs gpuva list.
+ *
+ * Returns: a pointer to the found &drm_gpuvm_bo or @__vm_bo if no existing
+ * &drm_gpuvm_bo was found
+ */
+struct drm_gpuvm_bo *
+drm_gpuvm_bo_obtain_prealloc(struct drm_gpuvm_bo *__vm_bo)
+{
+    struct drm_gpuvm *gpuvm = __vm_bo->vm;
+    struct drm_gem_object *obj = __vm_bo->obj;
+    struct drm_gpuvm_bo *vm_bo;
+
+    vm_bo = drm_gpuvm_bo_find(gpuvm, obj);
+    if (vm_bo) {
+        drm_gpuvm_bo_put(__vm_bo);
+        return vm_bo;
+    }
+
+    list_add_tail(&__vm_bo->list.entry.gem, &obj->gpuva.list);
+
+    return __vm_bo;
+}
+EXPORT_SYMBOL_GPL(drm_gpuvm_bo_obtain_prealloc);
+
  static int
  __drm_gpuva_insert(struct drm_gpuvm *gpuvm,
             struct drm_gpuva *va)
@@ -860,24 +1083,33 @@ EXPORT_SYMBOL_GPL(drm_gpuva_remove);
  /**
   * drm_gpuva_link() - link a &drm_gpuva
   * @va: the &drm_gpuva to link
+ * @vm_bo: the &drm_gpuvm_bo to add the &drm_gpuva to
   *
- * This adds the given &va to the GPU VA list of the &drm_gem_object it is
- * associated with.
+ * This adds the given &va to the GPU VA list of the &drm_gpuvm_bo and the
+ * &drm_gpuvm_bo to the &drm_gem_object it is associated with.
+ *
+ * For every &drm_gpuva entry added to the &drm_gpuvm_bo an additional
+ * reference of the latter is taken.
   *
   * This function expects the caller to protect the GEM's GPUVA list against
- * concurrent access using the GEMs dma_resv lock.
+ * concurrent access using either the GEMs dma_resv lock or a driver specific
+ * lock set through drm_gem_gpuva_set_lock().
   */
  void
-drm_gpuva_link(struct drm_gpuva *va)
+drm_gpuva_link(struct drm_gpuva *va, struct drm_gpuvm_bo *vm_bo)
  {
      struct drm_gem_object *obj = va->gem.obj;
      if (unlikely(!obj))
          return;
+    WARN_ON(obj != vm_bo->obj);
      drm_gem_gpuva_assert_lock_held(obj);
-    list_add_tail(&va->gem.entry, &obj->gpuva.list);
+    drm_gpuvm_bo_get(vm_bo);
+
+    va->vm_bo = vm_bo;
+    list_add_tail(&va->gem.entry, &vm_bo->list.gpuva);
  }
  EXPORT_SYMBOL_GPL(drm_gpuva_link);
@@ -888,13 +1120,22 @@ EXPORT_SYMBOL_GPL(drm_gpuva_link);
   * This removes the given &va from the GPU VA list of the &drm_gem_object it 
is
   * associated with.
   *
+ * This removes the given &va from the GPU VA list of the &drm_gpuvm_bo and
+ * the &drm_gpuvm_bo from the &drm_gem_object it is associated with in case
+ * this call unlinks the last &drm_gpuva from the &drm_gpuvm_bo.
+ *
+ * For every &drm_gpuva entry removed from the &drm_gpuvm_bo a reference of
+ * the latter is dropped.
+ *
   * This function expects the caller to protect the GEM's GPUVA list against
- * concurrent access using the GEMs dma_resv lock.
+ * concurrent access using either the GEMs dma_resv lock or a driver specific
+ * lock set through drm_gem_gpuva_set_lock().
   */
  void
  drm_gpuva_unlink(struct drm_gpuva *va)
  {
      struct drm_gem_object *obj = va->gem.obj;
Can we ditch va->gem.obj now and replace it with an accessor to the vm_bo's 
pointer?

Theoretically, drm_gpuvm could be used to track mappings and create 
drm_gpuva_ops to
map / unmap stuff only. Not sure if anyone ever does that though.

If we decide to drop it and use the vm_bo's pointer instead, the drm_gpuva must 
always
carry a pointer to the vm_bo since the drm_gpuvm_sm_map() function family 
requires the
BO pointer (if any) to work correctly. Also I'm not sure about all the 
corresponding
lifecycle implications yet.

Either way, I'd prefer to approach this in a separate patch if we decide so.

+    struct drm_gpuvm_bo *vm_bo = va->vm_bo;
      if (unlikely(!obj))
          return;
@@ -902,6 +1143,11 @@ drm_gpuva_unlink(struct drm_gpuva *va)
      drm_gem_gpuva_assert_lock_held(obj);
      list_del_init(&va->gem.entry);
+    va->vm_bo = NULL;
+
+    drm_gem_object_get(obj);
+    drm_gpuvm_bo_put(vm_bo);
+    drm_gem_object_put(obj);

This get->put dance is unneccesary? If the caller is required to hold a lock on 
obj it is also required to hold a reference on obj.

I think I had the case where the driver has a separate (external) GEM gpuva 
lock to
protect the GEM's VM_BO list and the VM_BO's drm_gpuva list in my mind when 
writing this.
However, in this case we also don't need to keep a reference on the GEM object 
I guess...

Will remove it.


Besides, if the vm_bo's reference on obj is otherwise the last one, it will 
still be freed before the function exits.

/Thomas



Reply via email to