[1/2] Enable buddy memory manager support

Commit Message

Paneer Selvam, Arunpravin Sept. 20, 2021, 7:20 p.m. UTC

Port Intel buddy system manager to drm root folder
Add CPU mappable/non-mappable region support to the drm buddy manager

Signed-off-by: Arunpravin <[e-mail]>
---
 drivers/gpu/drm/Makefile    |   2 +-
 drivers/gpu/drm/drm_buddy.c | 465 ++++++++++++++++++++++++++++++++++++
 include/drm/drm_buddy.h     | 154 ++++++++++++
 3 files changed, 620 insertions(+), 1 deletion(-)
 create mode 100644 drivers/gpu/drm/drm_buddy.c
 create mode 100644 include/drm/drm_buddy.h

Comments

Alex Deucher Sept. 20, 2021, 7:24 p.m. UTC | #1

On Mon, Sep 20, 2021 at 3:21 PM Arunpravin
<[e-mail]> wrote:

Please prefix the patch subject with drm.  E.g.,
drm: Enable buddy memory manager support

Same for the second patch, but make it drm/amdgpu instead.

Alex

>
> Port Intel buddy system manager to drm root folder
> Add CPU mappable/non-mappable region support to the drm buddy manager
>
> Signed-off-by: Arunpravin <[e-mail]>
> ---
>  drivers/gpu/drm/Makefile    |   2 +-
>  drivers/gpu/drm/drm_buddy.c | 465 ++++++++++++++++++++++++++++++++++++
>  include/drm/drm_buddy.h     | 154 ++++++++++++
>  3 files changed, 620 insertions(+), 1 deletion(-)
>  create mode 100644 drivers/gpu/drm/drm_buddy.c
>  create mode 100644 include/drm/drm_buddy.h
>
> diff --git a/drivers/gpu/drm/Makefile b/drivers/gpu/drm/Makefile
> index a118692a6df7..fe1a2fc09675 100644
> --- a/drivers/gpu/drm/Makefile
> +++ b/drivers/gpu/drm/Makefile
> @@ -18,7 +18,7 @@ drm-y       :=        drm_aperture.o drm_auth.o drm_cache.o \
>                 drm_dumb_buffers.o drm_mode_config.o drm_vblank.o \
>                 drm_syncobj.o drm_lease.o drm_writeback.o drm_client.o \
>                 drm_client_modeset.o drm_atomic_uapi.o drm_hdcp.o \
> -               drm_managed.o drm_vblank_work.o
> +               drm_managed.o drm_vblank_work.o drm_buddy.o
>
>  drm-$(CONFIG_DRM_LEGACY) += drm_agpsupport.o drm_bufs.o drm_context.o drm_dma.o \
>                             drm_legacy_misc.o drm_lock.o drm_memory.o drm_scatter.o \
> diff --git a/drivers/gpu/drm/drm_buddy.c b/drivers/gpu/drm/drm_buddy.c
> new file mode 100644
> index 000000000000..f07919a004b6
> --- /dev/null
> +++ b/drivers/gpu/drm/drm_buddy.c
> @@ -0,0 +1,465 @@
> +// SPDX-License-Identifier: MIT
> +/*
> + * Copyright � 2021 Intel Corporation
> + */
> +
> +#include <linux/kmemleak.h>
> +#include <drm/drm_buddy.h>
> +
> +static struct drm_buddy_block *drm_block_alloc(struct drm_buddy_mm *mm,
> +                       struct drm_buddy_block *parent, unsigned int order,
> +                       u64 offset)
> +{
> +       struct drm_buddy_block *block;
> +
> +       BUG_ON(order > DRM_BUDDY_MAX_ORDER);
> +
> +       block = kmem_cache_zalloc(mm->slab_blocks, GFP_KERNEL);
> +       if (!block)
> +               return NULL;
> +
> +       block->header = offset;
> +       block->header |= order;
> +       block->parent = parent;
> +       block->start = offset >> PAGE_SHIFT;
> +       block->size = (mm->chunk_size << order) >> PAGE_SHIFT;
> +
> +       BUG_ON(block->header & DRM_BUDDY_HEADER_UNUSED);
> +       return block;
> +}
> +
> +static void drm_block_free(struct drm_buddy_mm *mm, struct drm_buddy_block *block)
> +{
> +       kmem_cache_free(mm->slab_blocks, block);
> +}
> +
> +static void add_ordered(struct drm_buddy_mm *mm, struct drm_buddy_block *block)
> +{
> +       struct drm_buddy_block *node;
> +
> +       if (list_empty(&mm->free_list[drm_buddy_block_order(block)])) {
> +               list_add(&block->link,
> +                               &mm->free_list[drm_buddy_block_order(block)]);
> +               return;
> +       }
> +
> +       list_for_each_entry(node, &mm->free_list[drm_buddy_block_order(block)], link)
> +               if (block->start > node->start)
> +                       break;
> +
> +       __list_add(&block->link, node->link.prev, &node->link);
> +}
> +
> +static void mark_allocated(struct drm_buddy_block *block)
> +{
> +       block->header &= ~DRM_BUDDY_HEADER_STATE;
> +       block->header |= DRM_BUDDY_ALLOCATED;
> +
> +       list_del(&block->link);
> +}
> +
> +static void mark_free(struct drm_buddy_mm *mm,
> +                     struct drm_buddy_block *block)
> +{
> +       block->header &= ~DRM_BUDDY_HEADER_STATE;
> +       block->header |= DRM_BUDDY_FREE;
> +
> +       add_ordered(mm, block);
> +}
> +
> +static void mark_split(struct drm_buddy_block *block)
> +{
> +       block->header &= ~DRM_BUDDY_HEADER_STATE;
> +       block->header |= DRM_BUDDY_SPLIT;
> +
> +       list_del(&block->link);
> +}
> +
> +int drm_buddy_init(struct drm_buddy_mm *mm, u64 size, u64 chunk_size)
> +{
> +       unsigned int i;
> +       u64 offset;
> +
> +       if (size < chunk_size)
> +               return -EINVAL;
> +
> +       if (chunk_size < PAGE_SIZE)
> +               return -EINVAL;
> +
> +       if (!is_power_of_2(chunk_size))
> +               return -EINVAL;
> +
> +       size = round_down(size, chunk_size);
> +
> +       mm->size = size;
> +       mm->chunk_size = chunk_size;
> +       mm->max_order = ilog2(size) - ilog2(chunk_size);
> +
> +       BUG_ON(mm->max_order > DRM_BUDDY_MAX_ORDER);
> +
> +       mm->slab_blocks = KMEM_CACHE(drm_buddy_block, SLAB_HWCACHE_ALIGN);
> +
> +       if (!mm->slab_blocks)
> +               return -ENOMEM;
> +
> +       mm->free_list = kmalloc_array(mm->max_order + 1,
> +                                     sizeof(struct list_head),
> +                                     GFP_KERNEL);
> +       if (!mm->free_list)
> +               goto out_destroy_slab;
> +
> +       for (i = 0; i <= mm->max_order; ++i)
> +               INIT_LIST_HEAD(&mm->free_list[i]);
> +
> +       mm->n_roots = hweight64(size);
> +
> +       mm->roots = kmalloc_array(mm->n_roots,
> +                                 sizeof(struct drm_buddy_block *),
> +                                 GFP_KERNEL);
> +       if (!mm->roots)
> +               goto out_free_list;
> +
> +       offset = 0;
> +       i = 0;
> +
> +       /*
> +        * Split into power-of-two blocks, in case we are given a size that is
> +        * not itself a power-of-two.
> +        */
> +       do {
> +               struct drm_buddy_block *root;
> +               unsigned int order;
> +               u64 root_size;
> +
> +               root_size = rounddown_pow_of_two(size);
> +               order = ilog2(root_size) - ilog2(chunk_size);
> +
> +               root = drm_block_alloc(mm, NULL, order, offset);
> +               if (!root)
> +                       goto out_free_roots;
> +
> +               mark_free(mm, root);
> +
> +               BUG_ON(i > mm->max_order);
> +               BUG_ON(drm_buddy_block_size(mm, root) < chunk_size);
> +
> +               mm->roots[i] = root;
> +
> +               offset += root_size;
> +               size -= root_size;
> +               i++;
> +       } while (size);
> +
> +       return 0;
> +
> +out_free_roots:
> +       while (i--)
> +               drm_block_free(mm, mm->roots[i]);
> +       kfree(mm->roots);
> +out_free_list:
> +       kfree(mm->free_list);
> +out_destroy_slab:
> +       kmem_cache_destroy(mm->slab_blocks);
> +       return -ENOMEM;
> +}
> +EXPORT_SYMBOL(drm_buddy_init);
> +
> +void drm_buddy_fini(struct drm_buddy_mm *mm)
> +{
> +       int i;
> +
> +       for (i = 0; i < mm->n_roots; ++i) {
> +               WARN_ON(!drm_buddy_block_is_free(mm->roots[i]));
> +               drm_block_free(mm, mm->roots[i]);
> +       }
> +
> +       kfree(mm->roots);
> +       kfree(mm->free_list);
> +       kmem_cache_destroy(mm->slab_blocks);
> +}
> +EXPORT_SYMBOL(drm_buddy_fini);
> +
> +static int split_block(struct drm_buddy_mm *mm,
> +                      struct drm_buddy_block *block)
> +{
> +       unsigned int block_order = drm_buddy_block_order(block) - 1;
> +       u64 offset = drm_buddy_block_offset(block);
> +
> +       BUG_ON(!drm_buddy_block_is_free(block));
> +       BUG_ON(!drm_buddy_block_order(block));
> +
> +       block->left = drm_block_alloc(mm, block, block_order, offset);
> +       if (!block->left)
> +               return -ENOMEM;
> +
> +       block->right = drm_block_alloc(mm, block, block_order,
> +                                       offset + (mm->chunk_size << block_order));
> +       if (!block->right) {
> +               drm_block_free(mm, block->left);
> +               return -ENOMEM;
> +       }
> +
> +       mark_free(mm, block->left);
> +       mark_free(mm, block->right);
> +
> +       mark_split(block);
> +
> +       return 0;
> +}
> +
> +static struct drm_buddy_block *
> +get_buddy(struct drm_buddy_block *block)
> +{
> +       struct drm_buddy_block *parent;
> +
> +       parent = block->parent;
> +       if (!parent)
> +               return NULL;
> +
> +       if (parent->left == block)
> +               return parent->right;
> +
> +       return parent->left;
> +}
> +
> +static void __drm_buddy_free(struct drm_buddy_mm *mm,
> +                             struct drm_buddy_block *block)
> +{
> +       struct drm_buddy_block *parent;
> +
> +       while ((parent = block->parent)) {
> +               struct drm_buddy_block *buddy;
> +
> +               buddy = get_buddy(block);
> +
> +               if (!drm_buddy_block_is_free(buddy))
> +                       break;
> +
> +               list_del(&buddy->link);
> +
> +               drm_block_free(mm, block);
> +               drm_block_free(mm, buddy);
> +
> +               block = parent;
> +       }
> +
> +       mark_free(mm, block);
> +}
> +
> +void drm_buddy_free(struct drm_buddy_mm *mm,
> +                    struct drm_buddy_block *block)
> +{
> +       BUG_ON(!drm_buddy_block_is_allocated(block));
> +       __drm_buddy_free(mm, block);
> +}
> +EXPORT_SYMBOL(drm_buddy_free);
> +
> +void drm_buddy_free_list(struct drm_buddy_mm *mm, struct list_head *objects)
> +{
> +       struct drm_buddy_block *block, *on;
> +
> +       list_for_each_entry_safe(block, on, objects, link) {
> +               drm_buddy_free(mm, block);
> +               cond_resched();
> +       }
> +       INIT_LIST_HEAD(objects);
> +}
> +EXPORT_SYMBOL(drm_buddy_free_list);
> +
> +/*
> + * Allocate power-of-two block. The order value here translates to:
> + *
> + *   0 = 2^0 * mm->chunk_size
> + *   1 = 2^1 * mm->chunk_size
> + *   2 = 2^2 * mm->chunk_size
> + *   ...
> + */
> +struct drm_buddy_block *
> +drm_buddy_alloc(struct drm_buddy_mm *mm, unsigned int order,
> +               bool bar_enabled, unsigned int limit,
> +               enum drm_buddy_alloc_mode mode)
> +{
> +       struct drm_buddy_block *block = NULL;
> +       unsigned int pages;
> +       unsigned int i;
> +       int err;
> +
> +       pages = (mm->chunk_size << order) >> PAGE_SHIFT;
> +
> +       for (i = order; i <= mm->max_order; ++i) {
> +               if (mode == DRM_BUDDY_TOP_DOWN) {
> +                       if (!list_empty(&mm->free_list[i])) {
> +                               block = list_first_entry(&mm->free_list[i],
> +                                               struct drm_buddy_block, link);
> +
> +                               if (bar_enabled) {
> +                                       if (!(block->start > limit))
> +                                               continue;
> +                               }
> +
> +                               break;
> +                       }
> +               } else if (mode == DRM_BUDDY_BOTTOM_UP) {
> +                       if (!list_empty(&mm->free_list[i])) {
> +                               block = list_last_entry(&mm->free_list[i],
> +                                               struct drm_buddy_block, link);
> +
> +                               if (bar_enabled) {
> +                                       if (!(block->start < limit &&
> +                                                       (block->start + pages) < limit))
> +                                               continue;
> +                               }
> +
> +                               break;
> +                       }
> +               }
> +       }
> +
> +       if (!block)
> +               return ERR_PTR(-ENOSPC);
> +
> +       BUG_ON(!drm_buddy_block_is_free(block));
> +
> +       while (i != order) {
> +               err = split_block(mm, block);
> +               if (unlikely(err))
> +                       goto out_free;
> +
> +               /* Go low */
> +               if (mode == DRM_BUDDY_TOP_DOWN)
> +                       block = block->right;
> +               else
> +                       block = block->left;
> +               i--;
> +       }
> +
> +       if (mode == DRM_BUDDY_TOP_DOWN && bar_enabled) {
> +               if (!(block->start > limit))
> +                       return ERR_PTR(-ENOSPC);
> +       } else if (mode == DRM_BUDDY_BOTTOM_UP && bar_enabled) {
> +               if (!(block->start < limit &&
> +                               (block->start + pages) < limit))
> +                       return ERR_PTR(-ENOSPC);
> +       }
> +
> +       mark_allocated(block);
> +       kmemleak_update_trace(block);
> +       return block;
> +
> +out_free:
> +       if (i != order)
> +               __drm_buddy_free(mm, block);
> +       return ERR_PTR(err);
> +}
> +EXPORT_SYMBOL(drm_buddy_alloc);
> +
> +static inline bool overlaps(u64 s1, u64 e1, u64 s2, u64 e2)
> +{
> +       return s1 <= e2 && e1 >= s2;
> +}
> +
> +static inline bool contains(u64 s1, u64 e1, u64 s2, u64 e2)
> +{
> +       return s1 <= s2 && e1 >= e2;
> +}
> +
> +/*
> + * Allocate range. Note that it's safe to chain together multiple alloc_ranges
> + * with the same blocks list.
> + *
> + * Intended for pre-allocating portions of the address space, for example to
> + * reserve a block for the initial framebuffer or similar, hence the expectation
> + * here is that drm_buddy_alloc() is still the main vehicle for
> + * allocations, so if that's not the case then the drm_mm range allocator is
> + * probably a much better fit, and so you should probably go use that instead.
> + */
> +int drm_buddy_alloc_range(struct drm_buddy_mm *mm,
> +                       struct list_head *blocks,
> +                       u64 start, u64 size)
> +{
> +       struct drm_buddy_block *block;
> +       struct drm_buddy_block *buddy;
> +       LIST_HEAD(allocated);
> +       LIST_HEAD(dfs);
> +       u64 end;
> +       int err;
> +       int i;
> +
> +       if (size < mm->chunk_size)
> +               return -EINVAL;
> +
> +       if (!IS_ALIGNED(size | start, mm->chunk_size))
> +               return -EINVAL;
> +
> +       if (range_overflows(start, size, mm->size))
> +               return -EINVAL;
> +
> +       for (i = 0; i < mm->n_roots; ++i)
> +               list_add_tail(&mm->roots[i]->tmp_link, &dfs);
> +
> +       end = start + size - 1;
> +
> +       do {
> +               u64 block_start;
> +               u64 block_end;
> +
> +               block = list_first_entry_or_null(&dfs,
> +                                                struct drm_buddy_block,
> +                                                tmp_link);
> +               if (!block)
> +                       break;
> +
> +               list_del(&block->tmp_link);
> +
> +               block_start = drm_buddy_block_offset(block);
> +               block_end = block_start + drm_buddy_block_size(mm, block) - 1;
> +
> +               if (!overlaps(start, end, block_start, block_end))
> +                       continue;
> +
> +               if (drm_buddy_block_is_allocated(block)) {
> +                       err = -ENOSPC;
> +                       goto err_free;
> +               }
> +
> +               if (contains(start, end, block_start, block_end)) {
> +                       if (!drm_buddy_block_is_free(block)) {
> +                               err = -ENOSPC;
> +                               goto err_free;
> +                       }
> +
> +                       mark_allocated(block);
> +                       list_add_tail(&block->link, &allocated);
> +                       continue;
> +               }
> +
> +               if (!drm_buddy_block_is_split(block)) {
> +                       err = split_block(mm, block);
> +                       if (unlikely(err))
> +                               goto err_undo;
> +               }
> +
> +               list_add(&block->right->tmp_link, &dfs);
> +               list_add(&block->left->tmp_link, &dfs);
> +       } while (1);
> +
> +       list_splice_tail(&allocated, blocks);
> +       return 0;
> +
> +err_undo:
> +       /*
> +        * We really don't want to leave around a bunch of split blocks, since
> +        * bigger is better, so make sure we merge everything back before we
> +        * free the allocated blocks.
> +        */
> +       buddy = get_buddy(block);
> +       if (buddy &&
> +           (drm_buddy_block_is_free(block) &&
> +            drm_buddy_block_is_free(buddy)))
> +               __drm_buddy_free(mm, block);
> +
> +err_free:
> +       drm_buddy_free_list(mm, &allocated);
> +       return err;
> +}
> +EXPORT_SYMBOL(drm_buddy_alloc_range);
> diff --git a/include/drm/drm_buddy.h b/include/drm/drm_buddy.h
> new file mode 100644
> index 000000000000..50e4d819a3dd
> --- /dev/null
> +++ b/include/drm/drm_buddy.h
> @@ -0,0 +1,154 @@
> +/* SPDX-License-Identifier: MIT */
> +/*
> + * Copyright � 2021 Intel Corporation
> + */
> +
> +#ifndef __DRM_BUDDY_H__
> +#define __DRM_BUDDY_H__
> +
> +#include <linux/bitops.h>
> +#include <linux/list.h>
> +#include <linux/slab.h>
> +#include <linux/sched.h>
> +
> +#define range_overflows(start, size, max) ({ \
> +       typeof(start) start__ = (start); \
> +       typeof(size) size__ = (size); \
> +       typeof(max) max__ = (max); \
> +       (void)(&start__ == &size__); \
> +       (void)(&start__ == &max__); \
> +       start__ >= max__ || size__ > max__ - start__; \
> +})
> +
> +enum drm_buddy_alloc_mode {
> +        DRM_BUDDY_TOP_DOWN = 0,
> +        DRM_BUDDY_BOTTOM_UP,
> +        DRM_BUDDY_ALLOC_RANGE
> +};
> +
> +struct drm_buddy_block {
> +#define DRM_BUDDY_HEADER_OFFSET GENMASK_ULL(63, 12)
> +#define DRM_BUDDY_HEADER_STATE  GENMASK_ULL(11, 10)
> +#define   DRM_BUDDY_ALLOCATED     (1 << 10)
> +#define   DRM_BUDDY_FREE          (2 << 10)
> +#define   DRM_BUDDY_SPLIT         (3 << 10)
> +/* Free to be used, if needed in the future */
> +#define DRM_BUDDY_HEADER_UNUSED GENMASK_ULL(9, 6)
> +#define DRM_BUDDY_HEADER_ORDER  GENMASK_ULL(5, 0)
> +       u64 header;
> +       /* Store start and size fields in pages */
> +       u64 start;
> +       u64 size;
> +
> +       struct drm_buddy_block *left;
> +       struct drm_buddy_block *right;
> +       struct drm_buddy_block *parent;
> +
> +       void *private; /* owned by creator */
> +
> +       /*
> +        * While the block is allocated by the user through drm_buddy_alloc*,
> +        * the user has ownership of the link, for example to maintain within
> +        * a list, if so desired. As soon as the block is freed with
> +        * drm_buddy_free* ownership is given back to the mm.
> +        */
> +       struct list_head link;
> +       struct list_head tmp_link;
> +};
> +
> +/* Order-zero must be at least PAGE_SIZE */
> +#define DRM_BUDDY_MAX_ORDER (63 - PAGE_SHIFT)
> +
> +/*
> + * Binary Buddy System.
> + *
> + * Locking should be handled by the user, a simple mutex around
> + * drm_buddy_alloc* and drm_buddy_free* should suffice.
> + */
> +struct drm_buddy_mm {
> +       struct kmem_cache *slab_blocks;
> +       /* Maintain a free list for each order. */
> +       struct list_head *free_list;
> +
> +       /*
> +        * Maintain explicit binary tree(s) to track the allocation of the
> +        * address space. This gives us a simple way of finding a buddy block
> +        * and performing the potentially recursive merge step when freeing a
> +        * block.  Nodes are either allocated or free, in which case they will
> +        * also exist on the respective free list.
> +        */
> +       struct drm_buddy_block **roots;
> +
> +       /*
> +        * Anything from here is public, and remains static for the lifetime of
> +        * the mm. Everything above is considered do-not-touch.
> +        */
> +       unsigned int n_roots;
> +       unsigned int max_order;
> +
> +       /* Must be at least PAGE_SIZE */
> +       u64 chunk_size;
> +       u64 size;
> +};
> +
> +static inline u64
> +drm_buddy_block_offset(struct drm_buddy_block *block)
> +{
> +       return block->header & DRM_BUDDY_HEADER_OFFSET;
> +}
> +
> +static inline unsigned int
> +drm_buddy_block_order(struct drm_buddy_block *block)
> +{
> +       return block->header & DRM_BUDDY_HEADER_ORDER;
> +}
> +
> +static inline unsigned int
> +drm_buddy_block_state(struct drm_buddy_block *block)
> +{
> +       return block->header & DRM_BUDDY_HEADER_STATE;
> +}
> +
> +static inline bool
> +drm_buddy_block_is_allocated(struct drm_buddy_block *block)
> +{
> +       return drm_buddy_block_state(block) == DRM_BUDDY_ALLOCATED;
> +}
> +
> +static inline bool
> +drm_buddy_block_is_free(struct drm_buddy_block *block)
> +{
> +       return drm_buddy_block_state(block) == DRM_BUDDY_FREE;
> +}
> +
> +static inline bool
> +drm_buddy_block_is_split(struct drm_buddy_block *block)
> +{
> +       return drm_buddy_block_state(block) == DRM_BUDDY_SPLIT;
> +}
> +
> +static inline u64
> +drm_buddy_block_size(struct drm_buddy_mm *mm,
> +                     struct drm_buddy_block *block)
> +{
> +       return mm->chunk_size << drm_buddy_block_order(block);
> +}
> +
> +int drm_buddy_init(struct drm_buddy_mm *mm, u64 size, u64 chunk_size);
> +
> +void drm_buddy_fini(struct drm_buddy_mm *mm);
> +
> +struct drm_buddy_block *
> +drm_buddy_alloc(struct drm_buddy_mm *mm, unsigned int order,
> +               bool bar_limit_enabled, unsigned int limit,
> +               enum drm_buddy_alloc_mode mode);
> +
> +int drm_buddy_alloc_range(struct drm_buddy_mm *mm,
> +                          struct list_head *blocks,
> +                          u64 start, u64 size);
> +
> +void drm_buddy_free(struct drm_buddy_mm *mm, struct drm_buddy_block *block);
> +
> +void drm_buddy_free_list(struct drm_buddy_mm *mm, struct list_head *objects);
> +
> +#endif
> --
> 2.25.1
>

Christian König Sept. 22, 2021, 6:58 a.m. UTC | #2

Am 20.09.21 um 21:20 schrieb Arunpravin:
> Port Intel buddy system manager to drm root folder
> Add CPU mappable/non-mappable region support to the drm buddy manager

And where is the patch to switch i915 and remove the Intel copy of this?

In general I think that every public function here needs a kerneldoc 
description what it is all about.

Regards,
Christian.

>
> Signed-off-by: Arunpravin <[e-mail]>
> ---
>   drivers/gpu/drm/Makefile    |   2 +-
>   drivers/gpu/drm/drm_buddy.c | 465 ++++++++++++++++++++++++++++++++++++
>   include/drm/drm_buddy.h     | 154 ++++++++++++
>   3 files changed, 620 insertions(+), 1 deletion(-)
>   create mode 100644 drivers/gpu/drm/drm_buddy.c
>   create mode 100644 include/drm/drm_buddy.h
>
> diff --git a/drivers/gpu/drm/Makefile b/drivers/gpu/drm/Makefile
> index a118692a6df7..fe1a2fc09675 100644
> --- a/drivers/gpu/drm/Makefile
> +++ b/drivers/gpu/drm/Makefile
> @@ -18,7 +18,7 @@ drm-y       :=	drm_aperture.o drm_auth.o drm_cache.o \
>   		drm_dumb_buffers.o drm_mode_config.o drm_vblank.o \
>   		drm_syncobj.o drm_lease.o drm_writeback.o drm_client.o \
>   		drm_client_modeset.o drm_atomic_uapi.o drm_hdcp.o \
> -		drm_managed.o drm_vblank_work.o
> +		drm_managed.o drm_vblank_work.o drm_buddy.o
>   
>   drm-$(CONFIG_DRM_LEGACY) += drm_agpsupport.o drm_bufs.o drm_context.o drm_dma.o \
>   			    drm_legacy_misc.o drm_lock.o drm_memory.o drm_scatter.o \
> diff --git a/drivers/gpu/drm/drm_buddy.c b/drivers/gpu/drm/drm_buddy.c
> new file mode 100644
> index 000000000000..f07919a004b6
> --- /dev/null
> +++ b/drivers/gpu/drm/drm_buddy.c
> @@ -0,0 +1,465 @@
> +// SPDX-License-Identifier: MIT
> +/*
> + * Copyright � 2021 Intel Corporation
> + */
> +
> +#include <linux/kmemleak.h>
> +#include <drm/drm_buddy.h>
> +
> +static struct drm_buddy_block *drm_block_alloc(struct drm_buddy_mm *mm,
> +			struct drm_buddy_block *parent, unsigned int order,
> +			u64 offset)
> +{
> +	struct drm_buddy_block *block;
> +
> +	BUG_ON(order > DRM_BUDDY_MAX_ORDER);
> +
> +	block = kmem_cache_zalloc(mm->slab_blocks, GFP_KERNEL);
> +	if (!block)
> +		return NULL;
> +
> +	block->header = offset;
> +	block->header |= order;
> +	block->parent = parent;
> +	block->start = offset >> PAGE_SHIFT;
> +	block->size = (mm->chunk_size << order) >> PAGE_SHIFT;
> +
> +	BUG_ON(block->header & DRM_BUDDY_HEADER_UNUSED);
> +	return block;
> +}
> +
> +static void drm_block_free(struct drm_buddy_mm *mm, struct drm_buddy_block *block)
> +{
> +	kmem_cache_free(mm->slab_blocks, block);
> +}
> +
> +static void add_ordered(struct drm_buddy_mm *mm, struct drm_buddy_block *block)
> +{
> +	struct drm_buddy_block *node;
> +
> +	if (list_empty(&mm->free_list[drm_buddy_block_order(block)])) {
> +		list_add(&block->link,
> +				&mm->free_list[drm_buddy_block_order(block)]);
> +		return;
> +	}
> +
> +	list_for_each_entry(node, &mm->free_list[drm_buddy_block_order(block)], link)
> +		if (block->start > node->start)
> +			break;
> +
> +	__list_add(&block->link, node->link.prev, &node->link);
> +}
> +
> +static void mark_allocated(struct drm_buddy_block *block)
> +{
> +	block->header &= ~DRM_BUDDY_HEADER_STATE;
> +	block->header |= DRM_BUDDY_ALLOCATED;
> +
> +	list_del(&block->link);
> +}
> +
> +static void mark_free(struct drm_buddy_mm *mm,
> +		      struct drm_buddy_block *block)
> +{
> +	block->header &= ~DRM_BUDDY_HEADER_STATE;
> +	block->header |= DRM_BUDDY_FREE;
> +
> +	add_ordered(mm, block);
> +}
> +
> +static void mark_split(struct drm_buddy_block *block)
> +{
> +	block->header &= ~DRM_BUDDY_HEADER_STATE;
> +	block->header |= DRM_BUDDY_SPLIT;
> +
> +	list_del(&block->link);
> +}
> +
> +int drm_buddy_init(struct drm_buddy_mm *mm, u64 size, u64 chunk_size)
> +{
> +	unsigned int i;
> +	u64 offset;
> +
> +	if (size < chunk_size)
> +		return -EINVAL;
> +
> +	if (chunk_size < PAGE_SIZE)
> +		return -EINVAL;
> +
> +	if (!is_power_of_2(chunk_size))
> +		return -EINVAL;
> +
> +	size = round_down(size, chunk_size);
> +
> +	mm->size = size;
> +	mm->chunk_size = chunk_size;
> +	mm->max_order = ilog2(size) - ilog2(chunk_size);
> +
> +	BUG_ON(mm->max_order > DRM_BUDDY_MAX_ORDER);
> +
> +	mm->slab_blocks = KMEM_CACHE(drm_buddy_block, SLAB_HWCACHE_ALIGN);
> +
> +	if (!mm->slab_blocks)
> +		return -ENOMEM;
> +
> +	mm->free_list = kmalloc_array(mm->max_order + 1,
> +				      sizeof(struct list_head),
> +				      GFP_KERNEL);
> +	if (!mm->free_list)
> +		goto out_destroy_slab;
> +
> +	for (i = 0; i <= mm->max_order; ++i)
> +		INIT_LIST_HEAD(&mm->free_list[i]);
> +
> +	mm->n_roots = hweight64(size);
> +
> +	mm->roots = kmalloc_array(mm->n_roots,
> +				  sizeof(struct drm_buddy_block *),
> +				  GFP_KERNEL);
> +	if (!mm->roots)
> +		goto out_free_list;
> +
> +	offset = 0;
> +	i = 0;
> +
> +	/*
> +	 * Split into power-of-two blocks, in case we are given a size that is
> +	 * not itself a power-of-two.
> +	 */
> +	do {
> +		struct drm_buddy_block *root;
> +		unsigned int order;
> +		u64 root_size;
> +
> +		root_size = rounddown_pow_of_two(size);
> +		order = ilog2(root_size) - ilog2(chunk_size);
> +
> +		root = drm_block_alloc(mm, NULL, order, offset);
> +		if (!root)
> +			goto out_free_roots;
> +
> +		mark_free(mm, root);
> +
> +		BUG_ON(i > mm->max_order);
> +		BUG_ON(drm_buddy_block_size(mm, root) < chunk_size);
> +
> +		mm->roots[i] = root;
> +
> +		offset += root_size;
> +		size -= root_size;
> +		i++;
> +	} while (size);
> +
> +	return 0;
> +
> +out_free_roots:
> +	while (i--)
> +		drm_block_free(mm, mm->roots[i]);
> +	kfree(mm->roots);
> +out_free_list:
> +	kfree(mm->free_list);
> +out_destroy_slab:
> +	kmem_cache_destroy(mm->slab_blocks);
> +	return -ENOMEM;
> +}
> +EXPORT_SYMBOL(drm_buddy_init);
> +
> +void drm_buddy_fini(struct drm_buddy_mm *mm)
> +{
> +	int i;
> +
> +	for (i = 0; i < mm->n_roots; ++i) {
> +		WARN_ON(!drm_buddy_block_is_free(mm->roots[i]));
> +		drm_block_free(mm, mm->roots[i]);
> +	}
> +
> +	kfree(mm->roots);
> +	kfree(mm->free_list);
> +	kmem_cache_destroy(mm->slab_blocks);
> +}
> +EXPORT_SYMBOL(drm_buddy_fini);
> +
> +static int split_block(struct drm_buddy_mm *mm,
> +		       struct drm_buddy_block *block)
> +{
> +	unsigned int block_order = drm_buddy_block_order(block) - 1;
> +	u64 offset = drm_buddy_block_offset(block);
> +
> +	BUG_ON(!drm_buddy_block_is_free(block));
> +	BUG_ON(!drm_buddy_block_order(block));
> +
> +	block->left = drm_block_alloc(mm, block, block_order, offset);
> +	if (!block->left)
> +		return -ENOMEM;
> +
> +	block->right = drm_block_alloc(mm, block, block_order,
> +					offset + (mm->chunk_size << block_order));
> +	if (!block->right) {
> +		drm_block_free(mm, block->left);
> +		return -ENOMEM;
> +	}
> +
> +	mark_free(mm, block->left);
> +	mark_free(mm, block->right);
> +
> +	mark_split(block);
> +
> +	return 0;
> +}
> +
> +static struct drm_buddy_block *
> +get_buddy(struct drm_buddy_block *block)
> +{
> +	struct drm_buddy_block *parent;
> +
> +	parent = block->parent;
> +	if (!parent)
> +		return NULL;
> +
> +	if (parent->left == block)
> +		return parent->right;
> +
> +	return parent->left;
> +}
> +
> +static void __drm_buddy_free(struct drm_buddy_mm *mm,
> +			      struct drm_buddy_block *block)
> +{
> +	struct drm_buddy_block *parent;
> +
> +	while ((parent = block->parent)) {
> +		struct drm_buddy_block *buddy;
> +
> +		buddy = get_buddy(block);
> +
> +		if (!drm_buddy_block_is_free(buddy))
> +			break;
> +
> +		list_del(&buddy->link);
> +
> +		drm_block_free(mm, block);
> +		drm_block_free(mm, buddy);
> +
> +		block = parent;
> +	}
> +
> +	mark_free(mm, block);
> +}
> +
> +void drm_buddy_free(struct drm_buddy_mm *mm,
> +		     struct drm_buddy_block *block)
> +{
> +	BUG_ON(!drm_buddy_block_is_allocated(block));
> +	__drm_buddy_free(mm, block);
> +}
> +EXPORT_SYMBOL(drm_buddy_free);
> +
> +void drm_buddy_free_list(struct drm_buddy_mm *mm, struct list_head *objects)
> +{
> +	struct drm_buddy_block *block, *on;
> +
> +	list_for_each_entry_safe(block, on, objects, link) {
> +		drm_buddy_free(mm, block);
> +		cond_resched();
> +	}
> +	INIT_LIST_HEAD(objects);
> +}
> +EXPORT_SYMBOL(drm_buddy_free_list);
> +
> +/*
> + * Allocate power-of-two block. The order value here translates to:
> + *
> + *   0 = 2^0 * mm->chunk_size
> + *   1 = 2^1 * mm->chunk_size
> + *   2 = 2^2 * mm->chunk_size
> + *   ...
> + */
> +struct drm_buddy_block *
> +drm_buddy_alloc(struct drm_buddy_mm *mm, unsigned int order,
> +		bool bar_enabled, unsigned int limit,
> +		enum drm_buddy_alloc_mode mode)
> +{
> +	struct drm_buddy_block *block = NULL;
> +	unsigned int pages;
> +	unsigned int i;
> +	int err;
> +
> +	pages = (mm->chunk_size << order) >> PAGE_SHIFT;
> +
> +	for (i = order; i <= mm->max_order; ++i) {
> +		if (mode == DRM_BUDDY_TOP_DOWN) {
> +			if (!list_empty(&mm->free_list[i])) {
> +				block = list_first_entry(&mm->free_list[i],
> +						struct drm_buddy_block, link);
> +
> +				if (bar_enabled) {
> +					if (!(block->start > limit))
> +						continue;
> +				}
> +
> +				break;
> +			}
> +		} else if (mode == DRM_BUDDY_BOTTOM_UP) {
> +			if (!list_empty(&mm->free_list[i])) {
> +				block = list_last_entry(&mm->free_list[i],
> +						struct drm_buddy_block, link);
> +
> +				if (bar_enabled) {
> +					if (!(block->start < limit &&
> +							(block->start + pages) < limit))
> +						continue;
> +				}
> +
> +				break;
> +			}
> +		}
> +	}
> +
> +	if (!block)
> +		return ERR_PTR(-ENOSPC);
> +
> +	BUG_ON(!drm_buddy_block_is_free(block));
> +
> +	while (i != order) {
> +		err = split_block(mm, block);
> +		if (unlikely(err))
> +			goto out_free;
> +
> +		/* Go low */
> +		if (mode == DRM_BUDDY_TOP_DOWN)
> +			block = block->right;
> +		else
> +			block = block->left;
> +		i--;
> +	}
> +
> +	if (mode == DRM_BUDDY_TOP_DOWN && bar_enabled) {
> +		if (!(block->start > limit))
> +			return ERR_PTR(-ENOSPC);
> +	} else if (mode == DRM_BUDDY_BOTTOM_UP && bar_enabled) {
> +		if (!(block->start < limit &&
> +				(block->start + pages) < limit))
> +			return ERR_PTR(-ENOSPC);
> +	}
> +
> +	mark_allocated(block);
> +	kmemleak_update_trace(block);
> +	return block;
> +
> +out_free:
> +	if (i != order)
> +		__drm_buddy_free(mm, block);
> +	return ERR_PTR(err);
> +}
> +EXPORT_SYMBOL(drm_buddy_alloc);
> +
> +static inline bool overlaps(u64 s1, u64 e1, u64 s2, u64 e2)
> +{
> +	return s1 <= e2 && e1 >= s2;
> +}
> +
> +static inline bool contains(u64 s1, u64 e1, u64 s2, u64 e2)
> +{
> +	return s1 <= s2 && e1 >= e2;
> +}
> +
> +/*
> + * Allocate range. Note that it's safe to chain together multiple alloc_ranges
> + * with the same blocks list.
> + *
> + * Intended for pre-allocating portions of the address space, for example to
> + * reserve a block for the initial framebuffer or similar, hence the expectation
> + * here is that drm_buddy_alloc() is still the main vehicle for
> + * allocations, so if that's not the case then the drm_mm range allocator is
> + * probably a much better fit, and so you should probably go use that instead.
> + */
> +int drm_buddy_alloc_range(struct drm_buddy_mm *mm,
> +			struct list_head *blocks,
> +			u64 start, u64 size)
> +{
> +	struct drm_buddy_block *block;
> +	struct drm_buddy_block *buddy;
> +	LIST_HEAD(allocated);
> +	LIST_HEAD(dfs);
> +	u64 end;
> +	int err;
> +	int i;
> +
> +	if (size < mm->chunk_size)
> +		return -EINVAL;
> +
> +	if (!IS_ALIGNED(size | start, mm->chunk_size))
> +		return -EINVAL;
> +
> +	if (range_overflows(start, size, mm->size))
> +		return -EINVAL;
> +
> +	for (i = 0; i < mm->n_roots; ++i)
> +		list_add_tail(&mm->roots[i]->tmp_link, &dfs);
> +
> +	end = start + size - 1;
> +
> +	do {
> +		u64 block_start;
> +		u64 block_end;
> +
> +		block = list_first_entry_or_null(&dfs,
> +						 struct drm_buddy_block,
> +						 tmp_link);
> +		if (!block)
> +			break;
> +
> +		list_del(&block->tmp_link);
> +
> +		block_start = drm_buddy_block_offset(block);
> +		block_end = block_start + drm_buddy_block_size(mm, block) - 1;
> +
> +		if (!overlaps(start, end, block_start, block_end))
> +			continue;
> +
> +		if (drm_buddy_block_is_allocated(block)) {
> +			err = -ENOSPC;
> +			goto err_free;
> +		}
> +
> +		if (contains(start, end, block_start, block_end)) {
> +			if (!drm_buddy_block_is_free(block)) {
> +				err = -ENOSPC;
> +				goto err_free;
> +			}
> +
> +			mark_allocated(block);
> +			list_add_tail(&block->link, &allocated);
> +			continue;
> +		}
> +
> +		if (!drm_buddy_block_is_split(block)) {
> +			err = split_block(mm, block);
> +			if (unlikely(err))
> +				goto err_undo;
> +		}
> +
> +		list_add(&block->right->tmp_link, &dfs);
> +		list_add(&block->left->tmp_link, &dfs);
> +	} while (1);
> +
> +	list_splice_tail(&allocated, blocks);
> +	return 0;
> +
> +err_undo:
> +	/*
> +	 * We really don't want to leave around a bunch of split blocks, since
> +	 * bigger is better, so make sure we merge everything back before we
> +	 * free the allocated blocks.
> +	 */
> +	buddy = get_buddy(block);
> +	if (buddy &&
> +	    (drm_buddy_block_is_free(block) &&
> +	     drm_buddy_block_is_free(buddy)))
> +		__drm_buddy_free(mm, block);
> +
> +err_free:
> +	drm_buddy_free_list(mm, &allocated);
> +	return err;
> +}
> +EXPORT_SYMBOL(drm_buddy_alloc_range);
> diff --git a/include/drm/drm_buddy.h b/include/drm/drm_buddy.h
> new file mode 100644
> index 000000000000..50e4d819a3dd
> --- /dev/null
> +++ b/include/drm/drm_buddy.h
> @@ -0,0 +1,154 @@
> +/* SPDX-License-Identifier: MIT */
> +/*
> + * Copyright � 2021 Intel Corporation
> + */
> +
> +#ifndef __DRM_BUDDY_H__
> +#define __DRM_BUDDY_H__
> +
> +#include <linux/bitops.h>
> +#include <linux/list.h>
> +#include <linux/slab.h>
> +#include <linux/sched.h>
> +
> +#define range_overflows(start, size, max) ({ \
> +	typeof(start) start__ = (start); \
> +	typeof(size) size__ = (size); \
> +	typeof(max) max__ = (max); \
> +	(void)(&start__ == &size__); \
> +	(void)(&start__ == &max__); \
> +	start__ >= max__ || size__ > max__ - start__; \
> +})
> +
> +enum drm_buddy_alloc_mode {
> +	 DRM_BUDDY_TOP_DOWN = 0,
> +	 DRM_BUDDY_BOTTOM_UP,
> +	 DRM_BUDDY_ALLOC_RANGE
> +};
> +
> +struct drm_buddy_block {
> +#define DRM_BUDDY_HEADER_OFFSET GENMASK_ULL(63, 12)
> +#define DRM_BUDDY_HEADER_STATE  GENMASK_ULL(11, 10)
> +#define   DRM_BUDDY_ALLOCATED	   (1 << 10)
> +#define   DRM_BUDDY_FREE	   (2 << 10)
> +#define   DRM_BUDDY_SPLIT	   (3 << 10)
> +/* Free to be used, if needed in the future */
> +#define DRM_BUDDY_HEADER_UNUSED GENMASK_ULL(9, 6)
> +#define DRM_BUDDY_HEADER_ORDER  GENMASK_ULL(5, 0)
> +	u64 header;
> +	/* Store start and size fields in pages */
> +	u64 start;
> +	u64 size;
> +
> +	struct drm_buddy_block *left;
> +	struct drm_buddy_block *right;
> +	struct drm_buddy_block *parent;
> +
> +	void *private; /* owned by creator */
> +
> +	/*
> +	 * While the block is allocated by the user through drm_buddy_alloc*,
> +	 * the user has ownership of the link, for example to maintain within
> +	 * a list, if so desired. As soon as the block is freed with
> +	 * drm_buddy_free* ownership is given back to the mm.
> +	 */
> +	struct list_head link;
> +	struct list_head tmp_link;
> +};
> +
> +/* Order-zero must be at least PAGE_SIZE */
> +#define DRM_BUDDY_MAX_ORDER (63 - PAGE_SHIFT)
> +
> +/*
> + * Binary Buddy System.
> + *
> + * Locking should be handled by the user, a simple mutex around
> + * drm_buddy_alloc* and drm_buddy_free* should suffice.
> + */
> +struct drm_buddy_mm {
> +	struct kmem_cache *slab_blocks;
> +	/* Maintain a free list for each order. */
> +	struct list_head *free_list;
> +
> +	/*
> +	 * Maintain explicit binary tree(s) to track the allocation of the
> +	 * address space. This gives us a simple way of finding a buddy block
> +	 * and performing the potentially recursive merge step when freeing a
> +	 * block.  Nodes are either allocated or free, in which case they will
> +	 * also exist on the respective free list.
> +	 */
> +	struct drm_buddy_block **roots;
> +
> +	/*
> +	 * Anything from here is public, and remains static for the lifetime of
> +	 * the mm. Everything above is considered do-not-touch.
> +	 */
> +	unsigned int n_roots;
> +	unsigned int max_order;
> +
> +	/* Must be at least PAGE_SIZE */
> +	u64 chunk_size;
> +	u64 size;
> +};
> +
> +static inline u64
> +drm_buddy_block_offset(struct drm_buddy_block *block)
> +{
> +	return block->header & DRM_BUDDY_HEADER_OFFSET;
> +}
> +
> +static inline unsigned int
> +drm_buddy_block_order(struct drm_buddy_block *block)
> +{
> +	return block->header & DRM_BUDDY_HEADER_ORDER;
> +}
> +
> +static inline unsigned int
> +drm_buddy_block_state(struct drm_buddy_block *block)
> +{
> +	return block->header & DRM_BUDDY_HEADER_STATE;
> +}
> +
> +static inline bool
> +drm_buddy_block_is_allocated(struct drm_buddy_block *block)
> +{
> +	return drm_buddy_block_state(block) == DRM_BUDDY_ALLOCATED;
> +}
> +
> +static inline bool
> +drm_buddy_block_is_free(struct drm_buddy_block *block)
> +{
> +	return drm_buddy_block_state(block) == DRM_BUDDY_FREE;
> +}
> +
> +static inline bool
> +drm_buddy_block_is_split(struct drm_buddy_block *block)
> +{
> +	return drm_buddy_block_state(block) == DRM_BUDDY_SPLIT;
> +}
> +
> +static inline u64
> +drm_buddy_block_size(struct drm_buddy_mm *mm,
> +		      struct drm_buddy_block *block)
> +{
> +	return mm->chunk_size << drm_buddy_block_order(block);
> +}
> +
> +int drm_buddy_init(struct drm_buddy_mm *mm, u64 size, u64 chunk_size);
> +
> +void drm_buddy_fini(struct drm_buddy_mm *mm);
> +
> +struct drm_buddy_block *
> +drm_buddy_alloc(struct drm_buddy_mm *mm, unsigned int order,
> +		bool bar_limit_enabled, unsigned int limit,
> +		enum drm_buddy_alloc_mode mode);
> +
> +int drm_buddy_alloc_range(struct drm_buddy_mm *mm,
> +			   struct list_head *blocks,
> +			   u64 start, u64 size);
> +
> +void drm_buddy_free(struct drm_buddy_mm *mm, struct drm_buddy_block *block);
> +
> +void drm_buddy_free_list(struct drm_buddy_mm *mm, struct list_head *objects);
> +
> +#endif

Paneer Selvam, Arunpravin Sept. 23, 2021, 5:55 a.m. UTC | #3

[AMD Public Use]

Hi Alex,
I will fix the name and send a document in my next version.

Thanks,
Arun
-----Original Message-----
From: Alex Deucher <[e-mail]> 
Sent: Tuesday, September 21, 2021 12:54 AM
To: Paneer Selvam, Arunpravin <[e-mail]>
Cc: Maling list - DRI developers <[e-mail]>; Intel Graphics Development <[e-mail]>; amd-gfx list <[e-mail]>; Koenig, Christian <[e-mail]>; Matthew Auld <[e-mail]>; Daniel Vetter <[e-mail]>; Deucher, Alexander <[e-mail]>
Subject: Re: [PATCH 1/2] Enable buddy memory manager support

On Mon, Sep 20, 2021 at 3:21 PM Arunpravin <[e-mail]> wrote:

Please prefix the patch subject with drm.  E.g.,
drm: Enable buddy memory manager support

Same for the second patch, but make it drm/amdgpu instead.

Alex

>
> Port Intel buddy system manager to drm root folder Add CPU 
> mappable/non-mappable region support to the drm buddy manager
>
> Signed-off-by: Arunpravin <[e-mail]>
> ---
>  drivers/gpu/drm/Makefile    |   2 +-
>  drivers/gpu/drm/drm_buddy.c | 465 ++++++++++++++++++++++++++++++++++++
>  include/drm/drm_buddy.h     | 154 ++++++++++++
>  3 files changed, 620 insertions(+), 1 deletion(-)  create mode 100644 
> drivers/gpu/drm/drm_buddy.c  create mode 100644 
> include/drm/drm_buddy.h
>
> diff --git a/drivers/gpu/drm/Makefile b/drivers/gpu/drm/Makefile index 
> a118692a6df7..fe1a2fc09675 100644
> --- a/drivers/gpu/drm/Makefile
> +++ b/drivers/gpu/drm/Makefile
> @@ -18,7 +18,7 @@ drm-y       :=        drm_aperture.o drm_auth.o drm_cache.o \
>                 drm_dumb_buffers.o drm_mode_config.o drm_vblank.o \
>                 drm_syncobj.o drm_lease.o drm_writeback.o drm_client.o \
>                 drm_client_modeset.o drm_atomic_uapi.o drm_hdcp.o \
> -               drm_managed.o drm_vblank_work.o
> +               drm_managed.o drm_vblank_work.o drm_buddy.o
>
>  drm-$(CONFIG_DRM_LEGACY) += drm_agpsupport.o drm_bufs.o drm_context.o drm_dma.o \
>                             drm_legacy_misc.o drm_lock.o drm_memory.o 
> drm_scatter.o \ diff --git a/drivers/gpu/drm/drm_buddy.c 
> b/drivers/gpu/drm/drm_buddy.c new file mode 100644 index 
> 000000000000..f07919a004b6
> --- /dev/null
> +++ b/drivers/gpu/drm/drm_buddy.c
> @@ -0,0 +1,465 @@
> +// SPDX-License-Identifier: MIT
> +/*
> + * Copyright   2021 Intel Corporation  */
> +
> +#include <linux/kmemleak.h>
> +#include <drm/drm_buddy.h>
> +
> +static struct drm_buddy_block *drm_block_alloc(struct drm_buddy_mm *mm,
> +                       struct drm_buddy_block *parent, unsigned int order,
> +                       u64 offset)
> +{
> +       struct drm_buddy_block *block;
> +
> +       BUG_ON(order > DRM_BUDDY_MAX_ORDER);
> +
> +       block = kmem_cache_zalloc(mm->slab_blocks, GFP_KERNEL);
> +       if (!block)
> +               return NULL;
> +
> +       block->header = offset;
> +       block->header |= order;
> +       block->parent = parent;
> +       block->start = offset >> PAGE_SHIFT;
> +       block->size = (mm->chunk_size << order) >> PAGE_SHIFT;
> +
> +       BUG_ON(block->header & DRM_BUDDY_HEADER_UNUSED);
> +       return block;
> +}
> +
> +static void drm_block_free(struct drm_buddy_mm *mm, struct 
> +drm_buddy_block *block) {
> +       kmem_cache_free(mm->slab_blocks, block); }
> +
> +static void add_ordered(struct drm_buddy_mm *mm, struct 
> +drm_buddy_block *block) {
> +       struct drm_buddy_block *node;
> +
> +       if (list_empty(&mm->free_list[drm_buddy_block_order(block)])) {
> +               list_add(&block->link,
> +                               &mm->free_list[drm_buddy_block_order(block)]);
> +               return;
> +       }
> +
> +       list_for_each_entry(node, &mm->free_list[drm_buddy_block_order(block)], link)
> +               if (block->start > node->start)
> +                       break;
> +
> +       __list_add(&block->link, node->link.prev, &node->link); }
> +
> +static void mark_allocated(struct drm_buddy_block *block) {
> +       block->header &= ~DRM_BUDDY_HEADER_STATE;
> +       block->header |= DRM_BUDDY_ALLOCATED;
> +
> +       list_del(&block->link);
> +}
> +
> +static void mark_free(struct drm_buddy_mm *mm,
> +                     struct drm_buddy_block *block) {
> +       block->header &= ~DRM_BUDDY_HEADER_STATE;
> +       block->header |= DRM_BUDDY_FREE;
> +
> +       add_ordered(mm, block);
> +}
> +
> +static void mark_split(struct drm_buddy_block *block) {
> +       block->header &= ~DRM_BUDDY_HEADER_STATE;
> +       block->header |= DRM_BUDDY_SPLIT;
> +
> +       list_del(&block->link);
> +}
> +
> +int drm_buddy_init(struct drm_buddy_mm *mm, u64 size, u64 chunk_size) 
> +{
> +       unsigned int i;
> +       u64 offset;
> +
> +       if (size < chunk_size)
> +               return -EINVAL;
> +
> +       if (chunk_size < PAGE_SIZE)
> +               return -EINVAL;
> +
> +       if (!is_power_of_2(chunk_size))
> +               return -EINVAL;
> +
> +       size = round_down(size, chunk_size);
> +
> +       mm->size = size;
> +       mm->chunk_size = chunk_size;
> +       mm->max_order = ilog2(size) - ilog2(chunk_size);
> +
> +       BUG_ON(mm->max_order > DRM_BUDDY_MAX_ORDER);
> +
> +       mm->slab_blocks = KMEM_CACHE(drm_buddy_block, 
> + SLAB_HWCACHE_ALIGN);
> +
> +       if (!mm->slab_blocks)
> +               return -ENOMEM;
> +
> +       mm->free_list = kmalloc_array(mm->max_order + 1,
> +                                     sizeof(struct list_head),
> +                                     GFP_KERNEL);
> +       if (!mm->free_list)
> +               goto out_destroy_slab;
> +
> +       for (i = 0; i <= mm->max_order; ++i)
> +               INIT_LIST_HEAD(&mm->free_list[i]);
> +
> +       mm->n_roots = hweight64(size);
> +
> +       mm->roots = kmalloc_array(mm->n_roots,
> +                                 sizeof(struct drm_buddy_block *),
> +                                 GFP_KERNEL);
> +       if (!mm->roots)
> +               goto out_free_list;
> +
> +       offset = 0;
> +       i = 0;
> +
> +       /*
> +        * Split into power-of-two blocks, in case we are given a size that is
> +        * not itself a power-of-two.
> +        */
> +       do {
> +               struct drm_buddy_block *root;
> +               unsigned int order;
> +               u64 root_size;
> +
> +               root_size = rounddown_pow_of_two(size);
> +               order = ilog2(root_size) - ilog2(chunk_size);
> +
> +               root = drm_block_alloc(mm, NULL, order, offset);
> +               if (!root)
> +                       goto out_free_roots;
> +
> +               mark_free(mm, root);
> +
> +               BUG_ON(i > mm->max_order);
> +               BUG_ON(drm_buddy_block_size(mm, root) < chunk_size);
> +
> +               mm->roots[i] = root;
> +
> +               offset += root_size;
> +               size -= root_size;
> +               i++;
> +       } while (size);
> +
> +       return 0;
> +
> +out_free_roots:
> +       while (i--)
> +               drm_block_free(mm, mm->roots[i]);
> +       kfree(mm->roots);
> +out_free_list:
> +       kfree(mm->free_list);
> +out_destroy_slab:
> +       kmem_cache_destroy(mm->slab_blocks);
> +       return -ENOMEM;
> +}
> +EXPORT_SYMBOL(drm_buddy_init);
> +
> +void drm_buddy_fini(struct drm_buddy_mm *mm) {
> +       int i;
> +
> +       for (i = 0; i < mm->n_roots; ++i) {
> +               WARN_ON(!drm_buddy_block_is_free(mm->roots[i]));
> +               drm_block_free(mm, mm->roots[i]);
> +       }
> +
> +       kfree(mm->roots);
> +       kfree(mm->free_list);
> +       kmem_cache_destroy(mm->slab_blocks);
> +}
> +EXPORT_SYMBOL(drm_buddy_fini);
> +
> +static int split_block(struct drm_buddy_mm *mm,
> +                      struct drm_buddy_block *block) {
> +       unsigned int block_order = drm_buddy_block_order(block) - 1;
> +       u64 offset = drm_buddy_block_offset(block);
> +
> +       BUG_ON(!drm_buddy_block_is_free(block));
> +       BUG_ON(!drm_buddy_block_order(block));
> +
> +       block->left = drm_block_alloc(mm, block, block_order, offset);
> +       if (!block->left)
> +               return -ENOMEM;
> +
> +       block->right = drm_block_alloc(mm, block, block_order,
> +                                       offset + (mm->chunk_size << block_order));
> +       if (!block->right) {
> +               drm_block_free(mm, block->left);
> +               return -ENOMEM;
> +       }
> +
> +       mark_free(mm, block->left);
> +       mark_free(mm, block->right);
> +
> +       mark_split(block);
> +
> +       return 0;
> +}
> +
> +static struct drm_buddy_block *
> +get_buddy(struct drm_buddy_block *block) {
> +       struct drm_buddy_block *parent;
> +
> +       parent = block->parent;
> +       if (!parent)
> +               return NULL;
> +
> +       if (parent->left == block)
> +               return parent->right;
> +
> +       return parent->left;
> +}
> +
> +static void __drm_buddy_free(struct drm_buddy_mm *mm,
> +                             struct drm_buddy_block *block) {
> +       struct drm_buddy_block *parent;
> +
> +       while ((parent = block->parent)) {
> +               struct drm_buddy_block *buddy;
> +
> +               buddy = get_buddy(block);
> +
> +               if (!drm_buddy_block_is_free(buddy))
> +                       break;
> +
> +               list_del(&buddy->link);
> +
> +               drm_block_free(mm, block);
> +               drm_block_free(mm, buddy);
> +
> +               block = parent;
> +       }
> +
> +       mark_free(mm, block);
> +}
> +
> +void drm_buddy_free(struct drm_buddy_mm *mm,
> +                    struct drm_buddy_block *block) {
> +       BUG_ON(!drm_buddy_block_is_allocated(block));
> +       __drm_buddy_free(mm, block);
> +}
> +EXPORT_SYMBOL(drm_buddy_free);
> +
> +void drm_buddy_free_list(struct drm_buddy_mm *mm, struct list_head 
> +*objects) {
> +       struct drm_buddy_block *block, *on;
> +
> +       list_for_each_entry_safe(block, on, objects, link) {
> +               drm_buddy_free(mm, block);
> +               cond_resched();
> +       }
> +       INIT_LIST_HEAD(objects);
> +}
> +EXPORT_SYMBOL(drm_buddy_free_list);
> +
> +/*
> + * Allocate power-of-two block. The order value here translates to:
> + *
> + *   0 = 2^0 * mm->chunk_size
> + *   1 = 2^1 * mm->chunk_size
> + *   2 = 2^2 * mm->chunk_size
> + *   ...
> + */
> +struct drm_buddy_block *
> +drm_buddy_alloc(struct drm_buddy_mm *mm, unsigned int order,
> +               bool bar_enabled, unsigned int limit,
> +               enum drm_buddy_alloc_mode mode) {
> +       struct drm_buddy_block *block = NULL;
> +       unsigned int pages;
> +       unsigned int i;
> +       int err;
> +
> +       pages = (mm->chunk_size << order) >> PAGE_SHIFT;
> +
> +       for (i = order; i <= mm->max_order; ++i) {
> +               if (mode == DRM_BUDDY_TOP_DOWN) {
> +                       if (!list_empty(&mm->free_list[i])) {
> +                               block = list_first_entry(&mm->free_list[i],
> +                                               struct 
> + drm_buddy_block, link);
> +
> +                               if (bar_enabled) {
> +                                       if (!(block->start > limit))
> +                                               continue;
> +                               }
> +
> +                               break;
> +                       }
> +               } else if (mode == DRM_BUDDY_BOTTOM_UP) {
> +                       if (!list_empty(&mm->free_list[i])) {
> +                               block = list_last_entry(&mm->free_list[i],
> +                                               struct 
> + drm_buddy_block, link);
> +
> +                               if (bar_enabled) {
> +                                       if (!(block->start < limit &&
> +                                                       (block->start + pages) < limit))
> +                                               continue;
> +                               }
> +
> +                               break;
> +                       }
> +               }
> +       }
> +
> +       if (!block)
> +               return ERR_PTR(-ENOSPC);
> +
> +       BUG_ON(!drm_buddy_block_is_free(block));
> +
> +       while (i != order) {
> +               err = split_block(mm, block);
> +               if (unlikely(err))
> +                       goto out_free;
> +
> +               /* Go low */
> +               if (mode == DRM_BUDDY_TOP_DOWN)
> +                       block = block->right;
> +               else
> +                       block = block->left;
> +               i--;
> +       }
> +
> +       if (mode == DRM_BUDDY_TOP_DOWN && bar_enabled) {
> +               if (!(block->start > limit))
> +                       return ERR_PTR(-ENOSPC);
> +       } else if (mode == DRM_BUDDY_BOTTOM_UP && bar_enabled) {
> +               if (!(block->start < limit &&
> +                               (block->start + pages) < limit))
> +                       return ERR_PTR(-ENOSPC);
> +       }
> +
> +       mark_allocated(block);
> +       kmemleak_update_trace(block);
> +       return block;
> +
> +out_free:
> +       if (i != order)
> +               __drm_buddy_free(mm, block);
> +       return ERR_PTR(err);
> +}
> +EXPORT_SYMBOL(drm_buddy_alloc);
> +
> +static inline bool overlaps(u64 s1, u64 e1, u64 s2, u64 e2) {
> +       return s1 <= e2 && e1 >= s2;
> +}
> +
> +static inline bool contains(u64 s1, u64 e1, u64 s2, u64 e2) {
> +       return s1 <= s2 && e1 >= e2;
> +}
> +
> +/*
> + * Allocate range. Note that it's safe to chain together multiple 
> +alloc_ranges
> + * with the same blocks list.
> + *
> + * Intended for pre-allocating portions of the address space, for 
> +example to
> + * reserve a block for the initial framebuffer or similar, hence the 
> +expectation
> + * here is that drm_buddy_alloc() is still the main vehicle for
> + * allocations, so if that's not the case then the drm_mm range 
> +allocator is
> + * probably a much better fit, and so you should probably go use that instead.
> + */
> +int drm_buddy_alloc_range(struct drm_buddy_mm *mm,
> +                       struct list_head *blocks,
> +                       u64 start, u64 size) {
> +       struct drm_buddy_block *block;
> +       struct drm_buddy_block *buddy;
> +       LIST_HEAD(allocated);
> +       LIST_HEAD(dfs);
> +       u64 end;
> +       int err;
> +       int i;
> +
> +       if (size < mm->chunk_size)
> +               return -EINVAL;
> +
> +       if (!IS_ALIGNED(size | start, mm->chunk_size))
> +               return -EINVAL;
> +
> +       if (range_overflows(start, size, mm->size))
> +               return -EINVAL;
> +
> +       for (i = 0; i < mm->n_roots; ++i)
> +               list_add_tail(&mm->roots[i]->tmp_link, &dfs);
> +
> +       end = start + size - 1;
> +
> +       do {
> +               u64 block_start;
> +               u64 block_end;
> +
> +               block = list_first_entry_or_null(&dfs,
> +                                                struct drm_buddy_block,
> +                                                tmp_link);
> +               if (!block)
> +                       break;
> +
> +               list_del(&block->tmp_link);
> +
> +               block_start = drm_buddy_block_offset(block);
> +               block_end = block_start + drm_buddy_block_size(mm, 
> + block) - 1;
> +
> +               if (!overlaps(start, end, block_start, block_end))
> +                       continue;
> +
> +               if (drm_buddy_block_is_allocated(block)) {
> +                       err = -ENOSPC;
> +                       goto err_free;
> +               }
> +
> +               if (contains(start, end, block_start, block_end)) {
> +                       if (!drm_buddy_block_is_free(block)) {
> +                               err = -ENOSPC;
> +                               goto err_free;
> +                       }
> +
> +                       mark_allocated(block);
> +                       list_add_tail(&block->link, &allocated);
> +                       continue;
> +               }
> +
> +               if (!drm_buddy_block_is_split(block)) {
> +                       err = split_block(mm, block);
> +                       if (unlikely(err))
> +                               goto err_undo;
> +               }
> +
> +               list_add(&block->right->tmp_link, &dfs);
> +               list_add(&block->left->tmp_link, &dfs);
> +       } while (1);
> +
> +       list_splice_tail(&allocated, blocks);
> +       return 0;
> +
> +err_undo:
> +       /*
> +        * We really don't want to leave around a bunch of split blocks, since
> +        * bigger is better, so make sure we merge everything back before we
> +        * free the allocated blocks.
> +        */
> +       buddy = get_buddy(block);
> +       if (buddy &&
> +           (drm_buddy_block_is_free(block) &&
> +            drm_buddy_block_is_free(buddy)))
> +               __drm_buddy_free(mm, block);
> +
> +err_free:
> +       drm_buddy_free_list(mm, &allocated);
> +       return err;
> +}
> +EXPORT_SYMBOL(drm_buddy_alloc_range);
> diff --git a/include/drm/drm_buddy.h b/include/drm/drm_buddy.h new 
> file mode 100644 index 000000000000..50e4d819a3dd
> --- /dev/null
> +++ b/include/drm/drm_buddy.h
> @@ -0,0 +1,154 @@
> +/* SPDX-License-Identifier: MIT */
> +/*
> + * Copyright   2021 Intel Corporation  */
> +
> +#ifndef __DRM_BUDDY_H__
> +#define __DRM_BUDDY_H__
> +
> +#include <linux/bitops.h>
> +#include <linux/list.h>
> +#include <linux/slab.h>
> +#include <linux/sched.h>
> +
> +#define range_overflows(start, size, max) ({ \
> +       typeof(start) start__ = (start); \
> +       typeof(size) size__ = (size); \
> +       typeof(max) max__ = (max); \
> +       (void)(&start__ == &size__); \
> +       (void)(&start__ == &max__); \
> +       start__ >= max__ || size__ > max__ - start__; \
> +})
> +
> +enum drm_buddy_alloc_mode {
> +        DRM_BUDDY_TOP_DOWN = 0,
> +        DRM_BUDDY_BOTTOM_UP,
> +        DRM_BUDDY_ALLOC_RANGE
> +};
> +
> +struct drm_buddy_block {
> +#define DRM_BUDDY_HEADER_OFFSET GENMASK_ULL(63, 12) #define 
> +DRM_BUDDY_HEADER_STATE  GENMASK_ULL(11, 10)
> +#define   DRM_BUDDY_ALLOCATED     (1 << 10)
> +#define   DRM_BUDDY_FREE          (2 << 10)
> +#define   DRM_BUDDY_SPLIT         (3 << 10)
> +/* Free to be used, if needed in the future */ #define 
> +DRM_BUDDY_HEADER_UNUSED GENMASK_ULL(9, 6) #define 
> +DRM_BUDDY_HEADER_ORDER  GENMASK_ULL(5, 0)
> +       u64 header;
> +       /* Store start and size fields in pages */
> +       u64 start;
> +       u64 size;
> +
> +       struct drm_buddy_block *left;
> +       struct drm_buddy_block *right;
> +       struct drm_buddy_block *parent;
> +
> +       void *private; /* owned by creator */
> +
> +       /*
> +        * While the block is allocated by the user through drm_buddy_alloc*,
> +        * the user has ownership of the link, for example to maintain within
> +        * a list, if so desired. As soon as the block is freed with
> +        * drm_buddy_free* ownership is given back to the mm.
> +        */
> +       struct list_head link;
> +       struct list_head tmp_link;
> +};
> +
> +/* Order-zero must be at least PAGE_SIZE */ #define 
> +DRM_BUDDY_MAX_ORDER (63 - PAGE_SHIFT)
> +
> +/*
> + * Binary Buddy System.
> + *
> + * Locking should be handled by the user, a simple mutex around
> + * drm_buddy_alloc* and drm_buddy_free* should suffice.
> + */
> +struct drm_buddy_mm {
> +       struct kmem_cache *slab_blocks;
> +       /* Maintain a free list for each order. */
> +       struct list_head *free_list;
> +
> +       /*
> +        * Maintain explicit binary tree(s) to track the allocation of the
> +        * address space. This gives us a simple way of finding a buddy block
> +        * and performing the potentially recursive merge step when freeing a
> +        * block.  Nodes are either allocated or free, in which case they will
> +        * also exist on the respective free list.
> +        */
> +       struct drm_buddy_block **roots;
> +
> +       /*
> +        * Anything from here is public, and remains static for the lifetime of
> +        * the mm. Everything above is considered do-not-touch.
> +        */
> +       unsigned int n_roots;
> +       unsigned int max_order;
> +
> +       /* Must be at least PAGE_SIZE */
> +       u64 chunk_size;
> +       u64 size;
> +};
> +
> +static inline u64
> +drm_buddy_block_offset(struct drm_buddy_block *block) {
> +       return block->header & DRM_BUDDY_HEADER_OFFSET; }
> +
> +static inline unsigned int
> +drm_buddy_block_order(struct drm_buddy_block *block) {
> +       return block->header & DRM_BUDDY_HEADER_ORDER; }
> +
> +static inline unsigned int
> +drm_buddy_block_state(struct drm_buddy_block *block) {
> +       return block->header & DRM_BUDDY_HEADER_STATE; }
> +
> +static inline bool
> +drm_buddy_block_is_allocated(struct drm_buddy_block *block) {
> +       return drm_buddy_block_state(block) == DRM_BUDDY_ALLOCATED; }
> +
> +static inline bool
> +drm_buddy_block_is_free(struct drm_buddy_block *block) {
> +       return drm_buddy_block_state(block) == DRM_BUDDY_FREE; }
> +
> +static inline bool
> +drm_buddy_block_is_split(struct drm_buddy_block *block) {
> +       return drm_buddy_block_state(block) == DRM_BUDDY_SPLIT; }
> +
> +static inline u64
> +drm_buddy_block_size(struct drm_buddy_mm *mm,
> +                     struct drm_buddy_block *block) {
> +       return mm->chunk_size << drm_buddy_block_order(block); }
> +
> +int drm_buddy_init(struct drm_buddy_mm *mm, u64 size, u64 
> +chunk_size);
> +
> +void drm_buddy_fini(struct drm_buddy_mm *mm);
> +
> +struct drm_buddy_block *
> +drm_buddy_alloc(struct drm_buddy_mm *mm, unsigned int order,
> +               bool bar_limit_enabled, unsigned int limit,
> +               enum drm_buddy_alloc_mode mode);
> +
> +int drm_buddy_alloc_range(struct drm_buddy_mm *mm,
> +                          struct list_head *blocks,
> +                          u64 start, u64 size);
> +
> +void drm_buddy_free(struct drm_buddy_mm *mm, struct drm_buddy_block 
> +*block);
> +
> +void drm_buddy_free_list(struct drm_buddy_mm *mm, struct list_head 
> +*objects);
> +
> +#endif
> --
> 2.25.1
>

Paneer Selvam, Arunpravin Sept. 23, 2021, 5:59 a.m. UTC | #4

[AMD Public Use]

Hi Christian,

> And where is the patch to switch i915 and remove the Intel copy of this?
Creating a patch for the switch.

> In general I think that every public function here needs a kerneldoc description what it is all about.
Making a kernel doc description for each public function

Thanks,
Arun
-----Original Message-----
From: Koenig, Christian <[e-mail]> 
Sent: Wednesday, September 22, 2021 12:28 PM
To: Paneer Selvam, Arunpravin <[e-mail]>; [e-mail]; [e-mail]; [e-mail]; [e-mail]; [e-mail]; Deucher, Alexander <[e-mail]>
Subject: Re: [PATCH 1/2] Enable buddy memory manager support

Am 20.09.21 um 21:20 schrieb Arunpravin:
> Port Intel buddy system manager to drm root folder Add CPU 
> mappable/non-mappable region support to the drm buddy manager

And where is the patch to switch i915 and remove the Intel copy of this?

In general I think that every public function here needs a kerneldoc description what it is all about.

Regards,
Christian.

>
> Signed-off-by: Arunpravin <[e-mail]>
> ---
>   drivers/gpu/drm/Makefile    |   2 +-
>   drivers/gpu/drm/drm_buddy.c | 465 ++++++++++++++++++++++++++++++++++++
>   include/drm/drm_buddy.h     | 154 ++++++++++++
>   3 files changed, 620 insertions(+), 1 deletion(-)
>   create mode 100644 drivers/gpu/drm/drm_buddy.c
>   create mode 100644 include/drm/drm_buddy.h
>
> diff --git a/drivers/gpu/drm/Makefile b/drivers/gpu/drm/Makefile index 
> a118692a6df7..fe1a2fc09675 100644
> --- a/drivers/gpu/drm/Makefile
> +++ b/drivers/gpu/drm/Makefile
> @@ -18,7 +18,7 @@ drm-y       :=	drm_aperture.o drm_auth.o drm_cache.o \
>   		drm_dumb_buffers.o drm_mode_config.o drm_vblank.o \
>   		drm_syncobj.o drm_lease.o drm_writeback.o drm_client.o \
>   		drm_client_modeset.o drm_atomic_uapi.o drm_hdcp.o \
> -		drm_managed.o drm_vblank_work.o
> +		drm_managed.o drm_vblank_work.o drm_buddy.o
>   
>   drm-$(CONFIG_DRM_LEGACY) += drm_agpsupport.o drm_bufs.o drm_context.o drm_dma.o \
>   			    drm_legacy_misc.o drm_lock.o drm_memory.o drm_scatter.o \ 
> diff --git a/drivers/gpu/drm/drm_buddy.c b/drivers/gpu/drm/drm_buddy.c 
> new file mode 100644 index 000000000000..f07919a004b6
> --- /dev/null
> +++ b/drivers/gpu/drm/drm_buddy.c
> @@ -0,0 +1,465 @@
> +// SPDX-License-Identifier: MIT
> +/*
> + * Copyright   2021 Intel Corporation  */
> +
> +#include <linux/kmemleak.h>
> +#include <drm/drm_buddy.h>
> +
> +static struct drm_buddy_block *drm_block_alloc(struct drm_buddy_mm *mm,
> +			struct drm_buddy_block *parent, unsigned int order,
> +			u64 offset)
> +{
> +	struct drm_buddy_block *block;
> +
> +	BUG_ON(order > DRM_BUDDY_MAX_ORDER);
> +
> +	block = kmem_cache_zalloc(mm->slab_blocks, GFP_KERNEL);
> +	if (!block)
> +		return NULL;
> +
> +	block->header = offset;
> +	block->header |= order;
> +	block->parent = parent;
> +	block->start = offset >> PAGE_SHIFT;
> +	block->size = (mm->chunk_size << order) >> PAGE_SHIFT;
> +
> +	BUG_ON(block->header & DRM_BUDDY_HEADER_UNUSED);
> +	return block;
> +}
> +
> +static void drm_block_free(struct drm_buddy_mm *mm, struct 
> +drm_buddy_block *block) {
> +	kmem_cache_free(mm->slab_blocks, block); }
> +
> +static void add_ordered(struct drm_buddy_mm *mm, struct 
> +drm_buddy_block *block) {
> +	struct drm_buddy_block *node;
> +
> +	if (list_empty(&mm->free_list[drm_buddy_block_order(block)])) {
> +		list_add(&block->link,
> +				&mm->free_list[drm_buddy_block_order(block)]);
> +		return;
> +	}
> +
> +	list_for_each_entry(node, &mm->free_list[drm_buddy_block_order(block)], link)
> +		if (block->start > node->start)
> +			break;
> +
> +	__list_add(&block->link, node->link.prev, &node->link); }
> +
> +static void mark_allocated(struct drm_buddy_block *block) {
> +	block->header &= ~DRM_BUDDY_HEADER_STATE;
> +	block->header |= DRM_BUDDY_ALLOCATED;
> +
> +	list_del(&block->link);
> +}
> +
> +static void mark_free(struct drm_buddy_mm *mm,
> +		      struct drm_buddy_block *block) {
> +	block->header &= ~DRM_BUDDY_HEADER_STATE;
> +	block->header |= DRM_BUDDY_FREE;
> +
> +	add_ordered(mm, block);
> +}
> +
> +static void mark_split(struct drm_buddy_block *block) {
> +	block->header &= ~DRM_BUDDY_HEADER_STATE;
> +	block->header |= DRM_BUDDY_SPLIT;
> +
> +	list_del(&block->link);
> +}
> +
> +int drm_buddy_init(struct drm_buddy_mm *mm, u64 size, u64 chunk_size) 
> +{
> +	unsigned int i;
> +	u64 offset;
> +
> +	if (size < chunk_size)
> +		return -EINVAL;
> +
> +	if (chunk_size < PAGE_SIZE)
> +		return -EINVAL;
> +
> +	if (!is_power_of_2(chunk_size))
> +		return -EINVAL;
> +
> +	size = round_down(size, chunk_size);
> +
> +	mm->size = size;
> +	mm->chunk_size = chunk_size;
> +	mm->max_order = ilog2(size) - ilog2(chunk_size);
> +
> +	BUG_ON(mm->max_order > DRM_BUDDY_MAX_ORDER);
> +
> +	mm->slab_blocks = KMEM_CACHE(drm_buddy_block, SLAB_HWCACHE_ALIGN);
> +
> +	if (!mm->slab_blocks)
> +		return -ENOMEM;
> +
> +	mm->free_list = kmalloc_array(mm->max_order + 1,
> +				      sizeof(struct list_head),
> +				      GFP_KERNEL);
> +	if (!mm->free_list)
> +		goto out_destroy_slab;
> +
> +	for (i = 0; i <= mm->max_order; ++i)
> +		INIT_LIST_HEAD(&mm->free_list[i]);
> +
> +	mm->n_roots = hweight64(size);
> +
> +	mm->roots = kmalloc_array(mm->n_roots,
> +				  sizeof(struct drm_buddy_block *),
> +				  GFP_KERNEL);
> +	if (!mm->roots)
> +		goto out_free_list;
> +
> +	offset = 0;
> +	i = 0;
> +
> +	/*
> +	 * Split into power-of-two blocks, in case we are given a size that is
> +	 * not itself a power-of-two.
> +	 */
> +	do {
> +		struct drm_buddy_block *root;
> +		unsigned int order;
> +		u64 root_size;
> +
> +		root_size = rounddown_pow_of_two(size);
> +		order = ilog2(root_size) - ilog2(chunk_size);
> +
> +		root = drm_block_alloc(mm, NULL, order, offset);
> +		if (!root)
> +			goto out_free_roots;
> +
> +		mark_free(mm, root);
> +
> +		BUG_ON(i > mm->max_order);
> +		BUG_ON(drm_buddy_block_size(mm, root) < chunk_size);
> +
> +		mm->roots[i] = root;
> +
> +		offset += root_size;
> +		size -= root_size;
> +		i++;
> +	} while (size);
> +
> +	return 0;
> +
> +out_free_roots:
> +	while (i--)
> +		drm_block_free(mm, mm->roots[i]);
> +	kfree(mm->roots);
> +out_free_list:
> +	kfree(mm->free_list);
> +out_destroy_slab:
> +	kmem_cache_destroy(mm->slab_blocks);
> +	return -ENOMEM;
> +}
> +EXPORT_SYMBOL(drm_buddy_init);
> +
> +void drm_buddy_fini(struct drm_buddy_mm *mm) {
> +	int i;
> +
> +	for (i = 0; i < mm->n_roots; ++i) {
> +		WARN_ON(!drm_buddy_block_is_free(mm->roots[i]));
> +		drm_block_free(mm, mm->roots[i]);
> +	}
> +
> +	kfree(mm->roots);
> +	kfree(mm->free_list);
> +	kmem_cache_destroy(mm->slab_blocks);
> +}
> +EXPORT_SYMBOL(drm_buddy_fini);
> +
> +static int split_block(struct drm_buddy_mm *mm,
> +		       struct drm_buddy_block *block) {
> +	unsigned int block_order = drm_buddy_block_order(block) - 1;
> +	u64 offset = drm_buddy_block_offset(block);
> +
> +	BUG_ON(!drm_buddy_block_is_free(block));
> +	BUG_ON(!drm_buddy_block_order(block));
> +
> +	block->left = drm_block_alloc(mm, block, block_order, offset);
> +	if (!block->left)
> +		return -ENOMEM;
> +
> +	block->right = drm_block_alloc(mm, block, block_order,
> +					offset + (mm->chunk_size << block_order));
> +	if (!block->right) {
> +		drm_block_free(mm, block->left);
> +		return -ENOMEM;
> +	}
> +
> +	mark_free(mm, block->left);
> +	mark_free(mm, block->right);
> +
> +	mark_split(block);
> +
> +	return 0;
> +}
> +
> +static struct drm_buddy_block *
> +get_buddy(struct drm_buddy_block *block) {
> +	struct drm_buddy_block *parent;
> +
> +	parent = block->parent;
> +	if (!parent)
> +		return NULL;
> +
> +	if (parent->left == block)
> +		return parent->right;
> +
> +	return parent->left;
> +}
> +
> +static void __drm_buddy_free(struct drm_buddy_mm *mm,
> +			      struct drm_buddy_block *block) {
> +	struct drm_buddy_block *parent;
> +
> +	while ((parent = block->parent)) {
> +		struct drm_buddy_block *buddy;
> +
> +		buddy = get_buddy(block);
> +
> +		if (!drm_buddy_block_is_free(buddy))
> +			break;
> +
> +		list_del(&buddy->link);
> +
> +		drm_block_free(mm, block);
> +		drm_block_free(mm, buddy);
> +
> +		block = parent;
> +	}
> +
> +	mark_free(mm, block);
> +}
> +
> +void drm_buddy_free(struct drm_buddy_mm *mm,
> +		     struct drm_buddy_block *block) {
> +	BUG_ON(!drm_buddy_block_is_allocated(block));
> +	__drm_buddy_free(mm, block);
> +}
> +EXPORT_SYMBOL(drm_buddy_free);
> +
> +void drm_buddy_free_list(struct drm_buddy_mm *mm, struct list_head 
> +*objects) {
> +	struct drm_buddy_block *block, *on;
> +
> +	list_for_each_entry_safe(block, on, objects, link) {
> +		drm_buddy_free(mm, block);
> +		cond_resched();
> +	}
> +	INIT_LIST_HEAD(objects);
> +}
> +EXPORT_SYMBOL(drm_buddy_free_list);
> +
> +/*
> + * Allocate power-of-two block. The order value here translates to:
> + *
> + *   0 = 2^0 * mm->chunk_size
> + *   1 = 2^1 * mm->chunk_size
> + *   2 = 2^2 * mm->chunk_size
> + *   ...
> + */
> +struct drm_buddy_block *
> +drm_buddy_alloc(struct drm_buddy_mm *mm, unsigned int order,
> +		bool bar_enabled, unsigned int limit,
> +		enum drm_buddy_alloc_mode mode)
> +{
> +	struct drm_buddy_block *block = NULL;
> +	unsigned int pages;
> +	unsigned int i;
> +	int err;
> +
> +	pages = (mm->chunk_size << order) >> PAGE_SHIFT;
> +
> +	for (i = order; i <= mm->max_order; ++i) {
> +		if (mode == DRM_BUDDY_TOP_DOWN) {
> +			if (!list_empty(&mm->free_list[i])) {
> +				block = list_first_entry(&mm->free_list[i],
> +						struct drm_buddy_block, link);
> +
> +				if (bar_enabled) {
> +					if (!(block->start > limit))
> +						continue;
> +				}
> +
> +				break;
> +			}
> +		} else if (mode == DRM_BUDDY_BOTTOM_UP) {
> +			if (!list_empty(&mm->free_list[i])) {
> +				block = list_last_entry(&mm->free_list[i],
> +						struct drm_buddy_block, link);
> +
> +				if (bar_enabled) {
> +					if (!(block->start < limit &&
> +							(block->start + pages) < limit))
> +						continue;
> +				}
> +
> +				break;
> +			}
> +		}
> +	}
> +
> +	if (!block)
> +		return ERR_PTR(-ENOSPC);
> +
> +	BUG_ON(!drm_buddy_block_is_free(block));
> +
> +	while (i != order) {
> +		err = split_block(mm, block);
> +		if (unlikely(err))
> +			goto out_free;
> +
> +		/* Go low */
> +		if (mode == DRM_BUDDY_TOP_DOWN)
> +			block = block->right;
> +		else
> +			block = block->left;
> +		i--;
> +	}
> +
> +	if (mode == DRM_BUDDY_TOP_DOWN && bar_enabled) {
> +		if (!(block->start > limit))
> +			return ERR_PTR(-ENOSPC);
> +	} else if (mode == DRM_BUDDY_BOTTOM_UP && bar_enabled) {
> +		if (!(block->start < limit &&
> +				(block->start + pages) < limit))
> +			return ERR_PTR(-ENOSPC);
> +	}
> +
> +	mark_allocated(block);
> +	kmemleak_update_trace(block);
> +	return block;
> +
> +out_free:
> +	if (i != order)
> +		__drm_buddy_free(mm, block);
> +	return ERR_PTR(err);
> +}
> +EXPORT_SYMBOL(drm_buddy_alloc);
> +
> +static inline bool overlaps(u64 s1, u64 e1, u64 s2, u64 e2) {
> +	return s1 <= e2 && e1 >= s2;
> +}
> +
> +static inline bool contains(u64 s1, u64 e1, u64 s2, u64 e2) {
> +	return s1 <= s2 && e1 >= e2;
> +}
> +
> +/*
> + * Allocate range. Note that it's safe to chain together multiple 
> +alloc_ranges
> + * with the same blocks list.
> + *
> + * Intended for pre-allocating portions of the address space, for 
> +example to
> + * reserve a block for the initial framebuffer or similar, hence the 
> +expectation
> + * here is that drm_buddy_alloc() is still the main vehicle for
> + * allocations, so if that's not the case then the drm_mm range 
> +allocator is
> + * probably a much better fit, and so you should probably go use that instead.
> + */
> +int drm_buddy_alloc_range(struct drm_buddy_mm *mm,
> +			struct list_head *blocks,
> +			u64 start, u64 size)
> +{
> +	struct drm_buddy_block *block;
> +	struct drm_buddy_block *buddy;
> +	LIST_HEAD(allocated);
> +	LIST_HEAD(dfs);
> +	u64 end;
> +	int err;
> +	int i;
> +
> +	if (size < mm->chunk_size)
> +		return -EINVAL;
> +
> +	if (!IS_ALIGNED(size | start, mm->chunk_size))
> +		return -EINVAL;
> +
> +	if (range_overflows(start, size, mm->size))
> +		return -EINVAL;
> +
> +	for (i = 0; i < mm->n_roots; ++i)
> +		list_add_tail(&mm->roots[i]->tmp_link, &dfs);
> +
> +	end = start + size - 1;
> +
> +	do {
> +		u64 block_start;
> +		u64 block_end;
> +
> +		block = list_first_entry_or_null(&dfs,
> +						 struct drm_buddy_block,
> +						 tmp_link);
> +		if (!block)
> +			break;
> +
> +		list_del(&block->tmp_link);
> +
> +		block_start = drm_buddy_block_offset(block);
> +		block_end = block_start + drm_buddy_block_size(mm, block) - 1;
> +
> +		if (!overlaps(start, end, block_start, block_end))
> +			continue;
> +
> +		if (drm_buddy_block_is_allocated(block)) {
> +			err = -ENOSPC;
> +			goto err_free;
> +		}
> +
> +		if (contains(start, end, block_start, block_end)) {
> +			if (!drm_buddy_block_is_free(block)) {
> +				err = -ENOSPC;
> +				goto err_free;
> +			}
> +
> +			mark_allocated(block);
> +			list_add_tail(&block->link, &allocated);
> +			continue;
> +		}
> +
> +		if (!drm_buddy_block_is_split(block)) {
> +			err = split_block(mm, block);
> +			if (unlikely(err))
> +				goto err_undo;
> +		}
> +
> +		list_add(&block->right->tmp_link, &dfs);
> +		list_add(&block->left->tmp_link, &dfs);
> +	} while (1);
> +
> +	list_splice_tail(&allocated, blocks);
> +	return 0;
> +
> +err_undo:
> +	/*
> +	 * We really don't want to leave around a bunch of split blocks, since
> +	 * bigger is better, so make sure we merge everything back before we
> +	 * free the allocated blocks.
> +	 */
> +	buddy = get_buddy(block);
> +	if (buddy &&
> +	    (drm_buddy_block_is_free(block) &&
> +	     drm_buddy_block_is_free(buddy)))
> +		__drm_buddy_free(mm, block);
> +
> +err_free:
> +	drm_buddy_free_list(mm, &allocated);
> +	return err;
> +}
> +EXPORT_SYMBOL(drm_buddy_alloc_range);
> diff --git a/include/drm/drm_buddy.h b/include/drm/drm_buddy.h new 
> file mode 100644 index 000000000000..50e4d819a3dd
> --- /dev/null
> +++ b/include/drm/drm_buddy.h
> @@ -0,0 +1,154 @@
> +/* SPDX-License-Identifier: MIT */
> +/*
> + * Copyright   2021 Intel Corporation  */
> +
> +#ifndef __DRM_BUDDY_H__
> +#define __DRM_BUDDY_H__
> +
> +#include <linux/bitops.h>
> +#include <linux/list.h>
> +#include <linux/slab.h>
> +#include <linux/sched.h>
> +
> +#define range_overflows(start, size, max) ({ \
> +	typeof(start) start__ = (start); \
> +	typeof(size) size__ = (size); \
> +	typeof(max) max__ = (max); \
> +	(void)(&start__ == &size__); \
> +	(void)(&start__ == &max__); \
> +	start__ >= max__ || size__ > max__ - start__; \
> +})
> +
> +enum drm_buddy_alloc_mode {
> +	 DRM_BUDDY_TOP_DOWN = 0,
> +	 DRM_BUDDY_BOTTOM_UP,
> +	 DRM_BUDDY_ALLOC_RANGE
> +};
> +
> +struct drm_buddy_block {
> +#define DRM_BUDDY_HEADER_OFFSET GENMASK_ULL(63, 12) #define 
> +DRM_BUDDY_HEADER_STATE  GENMASK_ULL(11, 10)
> +#define   DRM_BUDDY_ALLOCATED	   (1 << 10)
> +#define   DRM_BUDDY_FREE	   (2 << 10)
> +#define   DRM_BUDDY_SPLIT	   (3 << 10)
> +/* Free to be used, if needed in the future */ #define 
> +DRM_BUDDY_HEADER_UNUSED GENMASK_ULL(9, 6) #define 
> +DRM_BUDDY_HEADER_ORDER  GENMASK_ULL(5, 0)
> +	u64 header;
> +	/* Store start and size fields in pages */
> +	u64 start;
> +	u64 size;
> +
> +	struct drm_buddy_block *left;
> +	struct drm_buddy_block *right;
> +	struct drm_buddy_block *parent;
> +
> +	void *private; /* owned by creator */
> +
> +	/*
> +	 * While the block is allocated by the user through drm_buddy_alloc*,
> +	 * the user has ownership of the link, for example to maintain within
> +	 * a list, if so desired. As soon as the block is freed with
> +	 * drm_buddy_free* ownership is given back to the mm.
> +	 */
> +	struct list_head link;
> +	struct list_head tmp_link;
> +};
> +
> +/* Order-zero must be at least PAGE_SIZE */ #define 
> +DRM_BUDDY_MAX_ORDER (63 - PAGE_SHIFT)
> +
> +/*
> + * Binary Buddy System.
> + *
> + * Locking should be handled by the user, a simple mutex around
> + * drm_buddy_alloc* and drm_buddy_free* should suffice.
> + */
> +struct drm_buddy_mm {
> +	struct kmem_cache *slab_blocks;
> +	/* Maintain a free list for each order. */
> +	struct list_head *free_list;
> +
> +	/*
> +	 * Maintain explicit binary tree(s) to track the allocation of the
> +	 * address space. This gives us a simple way of finding a buddy block
> +	 * and performing the potentially recursive merge step when freeing a
> +	 * block.  Nodes are either allocated or free, in which case they will
> +	 * also exist on the respective free list.
> +	 */
> +	struct drm_buddy_block **roots;
> +
> +	/*
> +	 * Anything from here is public, and remains static for the lifetime of
> +	 * the mm. Everything above is considered do-not-touch.
> +	 */
> +	unsigned int n_roots;
> +	unsigned int max_order;
> +
> +	/* Must be at least PAGE_SIZE */
> +	u64 chunk_size;
> +	u64 size;
> +};
> +
> +static inline u64
> +drm_buddy_block_offset(struct drm_buddy_block *block) {
> +	return block->header & DRM_BUDDY_HEADER_OFFSET; }
> +
> +static inline unsigned int
> +drm_buddy_block_order(struct drm_buddy_block *block) {
> +	return block->header & DRM_BUDDY_HEADER_ORDER; }
> +
> +static inline unsigned int
> +drm_buddy_block_state(struct drm_buddy_block *block) {
> +	return block->header & DRM_BUDDY_HEADER_STATE; }
> +
> +static inline bool
> +drm_buddy_block_is_allocated(struct drm_buddy_block *block) {
> +	return drm_buddy_block_state(block) == DRM_BUDDY_ALLOCATED; }
> +
> +static inline bool
> +drm_buddy_block_is_free(struct drm_buddy_block *block) {
> +	return drm_buddy_block_state(block) == DRM_BUDDY_FREE; }
> +
> +static inline bool
> +drm_buddy_block_is_split(struct drm_buddy_block *block) {
> +	return drm_buddy_block_state(block) == DRM_BUDDY_SPLIT; }
> +
> +static inline u64
> +drm_buddy_block_size(struct drm_buddy_mm *mm,
> +		      struct drm_buddy_block *block) {
> +	return mm->chunk_size << drm_buddy_block_order(block); }
> +
> +int drm_buddy_init(struct drm_buddy_mm *mm, u64 size, u64 
> +chunk_size);
> +
> +void drm_buddy_fini(struct drm_buddy_mm *mm);
> +
> +struct drm_buddy_block *
> +drm_buddy_alloc(struct drm_buddy_mm *mm, unsigned int order,
> +		bool bar_limit_enabled, unsigned int limit,
> +		enum drm_buddy_alloc_mode mode);
> +
> +int drm_buddy_alloc_range(struct drm_buddy_mm *mm,
> +			   struct list_head *blocks,
> +			   u64 start, u64 size);
> +
> +void drm_buddy_free(struct drm_buddy_mm *mm, struct drm_buddy_block 
> +*block);
> +
> +void drm_buddy_free_list(struct drm_buddy_mm *mm, struct list_head 
> +*objects);
> +
> +#endif

Patch

diff --git a/drivers/gpu/drm/Makefile b/drivers/gpu/drm/Makefile
index a118692a6df7..fe1a2fc09675 100644
--- a/drivers/gpu/drm/Makefile
+++ b/drivers/gpu/drm/Makefile
@@ -18,7 +18,7 @@  drm-y       :=	drm_aperture.o drm_auth.o drm_cache.o \
 		drm_dumb_buffers.o drm_mode_config.o drm_vblank.o \
 		drm_syncobj.o drm_lease.o drm_writeback.o drm_client.o \
 		drm_client_modeset.o drm_atomic_uapi.o drm_hdcp.o \
-		drm_managed.o drm_vblank_work.o
+		drm_managed.o drm_vblank_work.o drm_buddy.o
 
 drm-$(CONFIG_DRM_LEGACY) += drm_agpsupport.o drm_bufs.o drm_context.o drm_dma.o \
 			    drm_legacy_misc.o drm_lock.o drm_memory.o drm_scatter.o \
diff --git a/drivers/gpu/drm/drm_buddy.c b/drivers/gpu/drm/drm_buddy.c
new file mode 100644
index 000000000000..f07919a004b6
--- /dev/null
+++ b/drivers/gpu/drm/drm_buddy.c
@@ -0,0 +1,465 @@ 
+// SPDX-License-Identifier: MIT
+/*
+ * Copyright � 2021 Intel Corporation
+ */
+
+#include <linux/kmemleak.h>
+#include <drm/drm_buddy.h>
+
+static struct drm_buddy_block *drm_block_alloc(struct drm_buddy_mm *mm,
+			struct drm_buddy_block *parent, unsigned int order,
+			u64 offset)
+{
+	struct drm_buddy_block *block;
+
+	BUG_ON(order > DRM_BUDDY_MAX_ORDER);
+
+	block = kmem_cache_zalloc(mm->slab_blocks, GFP_KERNEL);
+	if (!block)
+		return NULL;
+
+	block->header = offset;
+	block->header |= order;
+	block->parent = parent;
+	block->start = offset >> PAGE_SHIFT;
+	block->size = (mm->chunk_size << order) >> PAGE_SHIFT;
+
+	BUG_ON(block->header & DRM_BUDDY_HEADER_UNUSED);
+	return block;
+}
+
+static void drm_block_free(struct drm_buddy_mm *mm, struct drm_buddy_block *block)
+{
+	kmem_cache_free(mm->slab_blocks, block);
+}
+
+static void add_ordered(struct drm_buddy_mm *mm, struct drm_buddy_block *block)
+{
+	struct drm_buddy_block *node;
+
+	if (list_empty(&mm->free_list[drm_buddy_block_order(block)])) {
+		list_add(&block->link,
+				&mm->free_list[drm_buddy_block_order(block)]);
+		return;
+	}
+
+	list_for_each_entry(node, &mm->free_list[drm_buddy_block_order(block)], link)
+		if (block->start > node->start)
+			break;
+
+	__list_add(&block->link, node->link.prev, &node->link);
+}
+
+static void mark_allocated(struct drm_buddy_block *block)
+{
+	block->header &= ~DRM_BUDDY_HEADER_STATE;
+	block->header |= DRM_BUDDY_ALLOCATED;
+
+	list_del(&block->link);
+}
+
+static void mark_free(struct drm_buddy_mm *mm,
+		      struct drm_buddy_block *block)
+{
+	block->header &= ~DRM_BUDDY_HEADER_STATE;
+	block->header |= DRM_BUDDY_FREE;
+
+	add_ordered(mm, block);
+}
+
+static void mark_split(struct drm_buddy_block *block)
+{
+	block->header &= ~DRM_BUDDY_HEADER_STATE;
+	block->header |= DRM_BUDDY_SPLIT;
+
+	list_del(&block->link);
+}
+
+int drm_buddy_init(struct drm_buddy_mm *mm, u64 size, u64 chunk_size)
+{
+	unsigned int i;
+	u64 offset;
+
+	if (size < chunk_size)
+		return -EINVAL;
+
+	if (chunk_size < PAGE_SIZE)
+		return -EINVAL;
+
+	if (!is_power_of_2(chunk_size))
+		return -EINVAL;
+
+	size = round_down(size, chunk_size);
+
+	mm->size = size;
+	mm->chunk_size = chunk_size;
+	mm->max_order = ilog2(size) - ilog2(chunk_size);
+
+	BUG_ON(mm->max_order > DRM_BUDDY_MAX_ORDER);
+
+	mm->slab_blocks = KMEM_CACHE(drm_buddy_block, SLAB_HWCACHE_ALIGN);
+
+	if (!mm->slab_blocks)
+		return -ENOMEM;
+
+	mm->free_list = kmalloc_array(mm->max_order + 1,
+				      sizeof(struct list_head),
+				      GFP_KERNEL);
+	if (!mm->free_list)
+		goto out_destroy_slab;
+
+	for (i = 0; i <= mm->max_order; ++i)
+		INIT_LIST_HEAD(&mm->free_list[i]);
+
+	mm->n_roots = hweight64(size);
+
+	mm->roots = kmalloc_array(mm->n_roots,
+				  sizeof(struct drm_buddy_block *),
+				  GFP_KERNEL);
+	if (!mm->roots)
+		goto out_free_list;
+
+	offset = 0;
+	i = 0;
+
+	/*
+	 * Split into power-of-two blocks, in case we are given a size that is
+	 * not itself a power-of-two.
+	 */
+	do {
+		struct drm_buddy_block *root;
+		unsigned int order;
+		u64 root_size;
+
+		root_size = rounddown_pow_of_two(size);
+		order = ilog2(root_size) - ilog2(chunk_size);
+
+		root = drm_block_alloc(mm, NULL, order, offset);
+		if (!root)
+			goto out_free_roots;
+
+		mark_free(mm, root);
+
+		BUG_ON(i > mm->max_order);
+		BUG_ON(drm_buddy_block_size(mm, root) < chunk_size);
+
+		mm->roots[i] = root;
+
+		offset += root_size;
+		size -= root_size;
+		i++;
+	} while (size);
+
+	return 0;
+
+out_free_roots:
+	while (i--)
+		drm_block_free(mm, mm->roots[i]);
+	kfree(mm->roots);
+out_free_list:
+	kfree(mm->free_list);
+out_destroy_slab:
+	kmem_cache_destroy(mm->slab_blocks);
+	return -ENOMEM;
+}
+EXPORT_SYMBOL(drm_buddy_init);
+
+void drm_buddy_fini(struct drm_buddy_mm *mm)
+{
+	int i;
+
+	for (i = 0; i < mm->n_roots; ++i) {
+		WARN_ON(!drm_buddy_block_is_free(mm->roots[i]));
+		drm_block_free(mm, mm->roots[i]);
+	}
+
+	kfree(mm->roots);
+	kfree(mm->free_list);
+	kmem_cache_destroy(mm->slab_blocks);
+}
+EXPORT_SYMBOL(drm_buddy_fini);
+
+static int split_block(struct drm_buddy_mm *mm,
+		       struct drm_buddy_block *block)
+{
+	unsigned int block_order = drm_buddy_block_order(block) - 1;
+	u64 offset = drm_buddy_block_offset(block);
+
+	BUG_ON(!drm_buddy_block_is_free(block));
+	BUG_ON(!drm_buddy_block_order(block));
+
+	block->left = drm_block_alloc(mm, block, block_order, offset);
+	if (!block->left)
+		return -ENOMEM;
+
+	block->right = drm_block_alloc(mm, block, block_order,
+					offset + (mm->chunk_size << block_order));
+	if (!block->right) {
+		drm_block_free(mm, block->left);
+		return -ENOMEM;
+	}
+
+	mark_free(mm, block->left);
+	mark_free(mm, block->right);
+
+	mark_split(block);
+
+	return 0;
+}
+
+static struct drm_buddy_block *
+get_buddy(struct drm_buddy_block *block)
+{
+	struct drm_buddy_block *parent;
+
+	parent = block->parent;
+	if (!parent)
+		return NULL;
+
+	if (parent->left == block)
+		return parent->right;
+
+	return parent->left;
+}
+
+static void __drm_buddy_free(struct drm_buddy_mm *mm,
+			      struct drm_buddy_block *block)
+{
+	struct drm_buddy_block *parent;
+
+	while ((parent = block->parent)) {
+		struct drm_buddy_block *buddy;
+
+		buddy = get_buddy(block);
+
+		if (!drm_buddy_block_is_free(buddy))
+			break;
+
+		list_del(&buddy->link);
+
+		drm_block_free(mm, block);
+		drm_block_free(mm, buddy);
+
+		block = parent;
+	}
+
+	mark_free(mm, block);
+}
+
+void drm_buddy_free(struct drm_buddy_mm *mm,
+		     struct drm_buddy_block *block)
+{
+	BUG_ON(!drm_buddy_block_is_allocated(block));
+	__drm_buddy_free(mm, block);
+}
+EXPORT_SYMBOL(drm_buddy_free);
+
+void drm_buddy_free_list(struct drm_buddy_mm *mm, struct list_head *objects)
+{
+	struct drm_buddy_block *block, *on;
+
+	list_for_each_entry_safe(block, on, objects, link) {
+		drm_buddy_free(mm, block);
+		cond_resched();
+	}
+	INIT_LIST_HEAD(objects);
+}
+EXPORT_SYMBOL(drm_buddy_free_list);
+
+/*
+ * Allocate power-of-two block. The order value here translates to:
+ *
+ *   0 = 2^0 * mm->chunk_size
+ *   1 = 2^1 * mm->chunk_size
+ *   2 = 2^2 * mm->chunk_size
+ *   ...
+ */
+struct drm_buddy_block *
+drm_buddy_alloc(struct drm_buddy_mm *mm, unsigned int order,
+		bool bar_enabled, unsigned int limit,
+		enum drm_buddy_alloc_mode mode)
+{
+	struct drm_buddy_block *block = NULL;
+	unsigned int pages;
+	unsigned int i;
+	int err;
+
+	pages = (mm->chunk_size << order) >> PAGE_SHIFT;
+
+	for (i = order; i <= mm->max_order; ++i) {
+		if (mode == DRM_BUDDY_TOP_DOWN) {
+			if (!list_empty(&mm->free_list[i])) {
+				block = list_first_entry(&mm->free_list[i],
+						struct drm_buddy_block, link);
+
+				if (bar_enabled) {
+					if (!(block->start > limit))
+						continue;
+				}
+
+				break;
+			}
+		} else if (mode == DRM_BUDDY_BOTTOM_UP) {
+			if (!list_empty(&mm->free_list[i])) {
+				block = list_last_entry(&mm->free_list[i],
+						struct drm_buddy_block, link);
+
+				if (bar_enabled) {
+					if (!(block->start < limit &&
+							(block->start + pages) < limit))
+						continue;
+				}
+
+				break;
+			}
+		}
+	}
+
+	if (!block)
+		return ERR_PTR(-ENOSPC);
+
+	BUG_ON(!drm_buddy_block_is_free(block));
+
+	while (i != order) {
+		err = split_block(mm, block);
+		if (unlikely(err))
+			goto out_free;
+
+		/* Go low */
+		if (mode == DRM_BUDDY_TOP_DOWN)
+			block = block->right;
+		else
+			block = block->left;
+		i--;
+	}
+
+	if (mode == DRM_BUDDY_TOP_DOWN && bar_enabled) {
+		if (!(block->start > limit))
+			return ERR_PTR(-ENOSPC);
+	} else if (mode == DRM_BUDDY_BOTTOM_UP && bar_enabled) {
+		if (!(block->start < limit &&
+				(block->start + pages) < limit))
+			return ERR_PTR(-ENOSPC);
+	}
+
+	mark_allocated(block);
+	kmemleak_update_trace(block);
+	return block;
+
+out_free:
+	if (i != order)
+		__drm_buddy_free(mm, block);
+	return ERR_PTR(err);
+}
+EXPORT_SYMBOL(drm_buddy_alloc);
+
+static inline bool overlaps(u64 s1, u64 e1, u64 s2, u64 e2)
+{
+	return s1 <= e2 && e1 >= s2;
+}
+
+static inline bool contains(u64 s1, u64 e1, u64 s2, u64 e2)
+{
+	return s1 <= s2 && e1 >= e2;
+}
+
+/*
+ * Allocate range. Note that it's safe to chain together multiple alloc_ranges
+ * with the same blocks list.
+ *
+ * Intended for pre-allocating portions of the address space, for example to
+ * reserve a block for the initial framebuffer or similar, hence the expectation
+ * here is that drm_buddy_alloc() is still the main vehicle for
+ * allocations, so if that's not the case then the drm_mm range allocator is
+ * probably a much better fit, and so you should probably go use that instead.
+ */
+int drm_buddy_alloc_range(struct drm_buddy_mm *mm,
+			struct list_head *blocks,
+			u64 start, u64 size)
+{
+	struct drm_buddy_block *block;
+	struct drm_buddy_block *buddy;
+	LIST_HEAD(allocated);
+	LIST_HEAD(dfs);
+	u64 end;
+	int err;
+	int i;
+
+	if (size < mm->chunk_size)
+		return -EINVAL;
+
+	if (!IS_ALIGNED(size | start, mm->chunk_size))
+		return -EINVAL;
+
+	if (range_overflows(start, size, mm->size))
+		return -EINVAL;
+
+	for (i = 0; i < mm->n_roots; ++i)
+		list_add_tail(&mm->roots[i]->tmp_link, &dfs);
+
+	end = start + size - 1;
+
+	do {
+		u64 block_start;
+		u64 block_end;
+
+		block = list_first_entry_or_null(&dfs,
+						 struct drm_buddy_block,
+						 tmp_link);
+		if (!block)
+			break;
+
+		list_del(&block->tmp_link);
+
+		block_start = drm_buddy_block_offset(block);
+		block_end = block_start + drm_buddy_block_size(mm, block) - 1;
+
+		if (!overlaps(start, end, block_start, block_end))
+			continue;
+
+		if (drm_buddy_block_is_allocated(block)) {
+			err = -ENOSPC;
+			goto err_free;
+		}
+
+		if (contains(start, end, block_start, block_end)) {
+			if (!drm_buddy_block_is_free(block)) {
+				err = -ENOSPC;
+				goto err_free;
+			}
+
+			mark_allocated(block);
+			list_add_tail(&block->link, &allocated);
+			continue;
+		}
+
+		if (!drm_buddy_block_is_split(block)) {
+			err = split_block(mm, block);
+			if (unlikely(err))
+				goto err_undo;
+		}
+
+		list_add(&block->right->tmp_link, &dfs);
+		list_add(&block->left->tmp_link, &dfs);
+	} while (1);
+
+	list_splice_tail(&allocated, blocks);
+	return 0;
+
+err_undo:
+	/*
+	 * We really don't want to leave around a bunch of split blocks, since
+	 * bigger is better, so make sure we merge everything back before we
+	 * free the allocated blocks.
+	 */
+	buddy = get_buddy(block);
+	if (buddy &&
+	    (drm_buddy_block_is_free(block) &&
+	     drm_buddy_block_is_free(buddy)))
+		__drm_buddy_free(mm, block);
+
+err_free:
+	drm_buddy_free_list(mm, &allocated);
+	return err;
+}
+EXPORT_SYMBOL(drm_buddy_alloc_range);
diff --git a/include/drm/drm_buddy.h b/include/drm/drm_buddy.h
new file mode 100644
index 000000000000..50e4d819a3dd
--- /dev/null
+++ b/include/drm/drm_buddy.h
@@ -0,0 +1,154 @@ 
+/* SPDX-License-Identifier: MIT */
+/*
+ * Copyright � 2021 Intel Corporation
+ */
+
+#ifndef __DRM_BUDDY_H__
+#define __DRM_BUDDY_H__
+
+#include <linux/bitops.h>
+#include <linux/list.h>
+#include <linux/slab.h>
+#include <linux/sched.h>
+
+#define range_overflows(start, size, max) ({ \
+	typeof(start) start__ = (start); \
+	typeof(size) size__ = (size); \
+	typeof(max) max__ = (max); \
+	(void)(&start__ == &size__); \
+	(void)(&start__ == &max__); \
+	start__ >= max__ || size__ > max__ - start__; \
+})
+
+enum drm_buddy_alloc_mode {
+	 DRM_BUDDY_TOP_DOWN = 0,
+	 DRM_BUDDY_BOTTOM_UP,
+	 DRM_BUDDY_ALLOC_RANGE
+};
+
+struct drm_buddy_block {
+#define DRM_BUDDY_HEADER_OFFSET GENMASK_ULL(63, 12)
+#define DRM_BUDDY_HEADER_STATE  GENMASK_ULL(11, 10)
+#define   DRM_BUDDY_ALLOCATED	   (1 << 10)
+#define   DRM_BUDDY_FREE	   (2 << 10)
+#define   DRM_BUDDY_SPLIT	   (3 << 10)
+/* Free to be used, if needed in the future */
+#define DRM_BUDDY_HEADER_UNUSED GENMASK_ULL(9, 6)
+#define DRM_BUDDY_HEADER_ORDER  GENMASK_ULL(5, 0)
+	u64 header;
+	/* Store start and size fields in pages */
+	u64 start;
+	u64 size;
+
+	struct drm_buddy_block *left;
+	struct drm_buddy_block *right;
+	struct drm_buddy_block *parent;
+
+	void *private; /* owned by creator */
+
+	/*
+	 * While the block is allocated by the user through drm_buddy_alloc*,
+	 * the user has ownership of the link, for example to maintain within
+	 * a list, if so desired. As soon as the block is freed with
+	 * drm_buddy_free* ownership is given back to the mm.
+	 */
+	struct list_head link;
+	struct list_head tmp_link;
+};
+
+/* Order-zero must be at least PAGE_SIZE */
+#define DRM_BUDDY_MAX_ORDER (63 - PAGE_SHIFT)
+
+/*
+ * Binary Buddy System.
+ *
+ * Locking should be handled by the user, a simple mutex around
+ * drm_buddy_alloc* and drm_buddy_free* should suffice.
+ */
+struct drm_buddy_mm {
+	struct kmem_cache *slab_blocks;
+	/* Maintain a free list for each order. */
+	struct list_head *free_list;
+
+	/*
+	 * Maintain explicit binary tree(s) to track the allocation of the
+	 * address space. This gives us a simple way of finding a buddy block
+	 * and performing the potentially recursive merge step when freeing a
+	 * block.  Nodes are either allocated or free, in which case they will
+	 * also exist on the respective free list.
+	 */
+	struct drm_buddy_block **roots;
+
+	/*
+	 * Anything from here is public, and remains static for the lifetime of
+	 * the mm. Everything above is considered do-not-touch.
+	 */
+	unsigned int n_roots;
+	unsigned int max_order;
+
+	/* Must be at least PAGE_SIZE */
+	u64 chunk_size;
+	u64 size;
+};
+
+static inline u64
+drm_buddy_block_offset(struct drm_buddy_block *block)
+{
+	return block->header & DRM_BUDDY_HEADER_OFFSET;
+}
+
+static inline unsigned int
+drm_buddy_block_order(struct drm_buddy_block *block)
+{
+	return block->header & DRM_BUDDY_HEADER_ORDER;
+}
+
+static inline unsigned int
+drm_buddy_block_state(struct drm_buddy_block *block)
+{
+	return block->header & DRM_BUDDY_HEADER_STATE;
+}
+
+static inline bool
+drm_buddy_block_is_allocated(struct drm_buddy_block *block)
+{
+	return drm_buddy_block_state(block) == DRM_BUDDY_ALLOCATED;
+}
+
+static inline bool
+drm_buddy_block_is_free(struct drm_buddy_block *block)
+{
+	return drm_buddy_block_state(block) == DRM_BUDDY_FREE;
+}
+
+static inline bool
+drm_buddy_block_is_split(struct drm_buddy_block *block)
+{
+	return drm_buddy_block_state(block) == DRM_BUDDY_SPLIT;
+}
+
+static inline u64
+drm_buddy_block_size(struct drm_buddy_mm *mm,
+		      struct drm_buddy_block *block)
+{
+	return mm->chunk_size << drm_buddy_block_order(block);
+}
+
+int drm_buddy_init(struct drm_buddy_mm *mm, u64 size, u64 chunk_size);
+
+void drm_buddy_fini(struct drm_buddy_mm *mm);
+
+struct drm_buddy_block *
+drm_buddy_alloc(struct drm_buddy_mm *mm, unsigned int order,
+		bool bar_limit_enabled, unsigned int limit,
+		enum drm_buddy_alloc_mode mode);
+
+int drm_buddy_alloc_range(struct drm_buddy_mm *mm,
+			   struct list_head *blocks,
+			   u64 start, u64 size);
+
+void drm_buddy_free(struct drm_buddy_mm *mm, struct drm_buddy_block *block);
+
+void drm_buddy_free_list(struct drm_buddy_mm *mm, struct list_head *objects);
+
+#endif